PDA

View Full Version : Canon 7D, meet - the Canon 7D



Barry_Green
10-15-2009, 09:34 AM
Okay, in an attempt to please all of the people all of the time, I have shot another still vs. movie test.

To mollify lens complaints, I used the Zeiss ZF 50mm. I assume nobody is going to complain about that? If so, I can't help you.

To mollify aperture complaints/diffraction, I shot at f/5.6.

To mollify noise complaints, I chose ISO 160, which I believe to be the cleanest ISO the camera offers.

To mollify "sharpness" complaints, I used the default Landscape mode. I shot a landscape, and I used the landscape mode, which defaults to sharpness 3. Any complaints about that, please direct them to www.canon.com (http://www.canon.com). Thank you.

To mollify "focus" complaints, I set focus with the iris at f/1.4, using the 10x magnification, and then stopped down to f/5.6. It's focused, okay?

To mollify any potential "bias" complaints, I eliminated all other products and am comparing the Canon against the Canon. If anyone can find any bias there, they really, really need to take a break from the boards and go get a different hobby.

To prevent any complaints about ND filters, I didn't use one. I used Av mode, and let the camera pick its own shutter speed, seeing as this is primarily meant to be a comparison to show how clean the image could be, vs. what the camera's internal hardware processing does to the image in movie mode, therefore I didn't consider motion a priority. Be aware, however, that you cannot achieve results like this in real movie shooting without some compromise (either to the shutter speed or using an ND filter or stopping down more; if you want to consider the effect of those, please see the other thread). Once the camera had suggested the shutter speed, I went back to fully manual mode and set the shutter and ISO manually, and of course the aperture was manual because it's a manual lens.

To prevent any complaints about any possibility of anything whatsoever being any different in any way at all, I shot the movie and then while shooting it, pressed the "still" button so I actually took a still during the movie. That should be as absolutely identical as it's possibly possible to be. Although, to be fair, the the still extracted from the video was picked at random.

To avoid anyone throwing in flare issues, I shot with the sun at my back.

Finally, these are not the exact same shot as where the car shot was taken. That's done on purpose, because I'd never be able to match exactly, plus I wasn't at 50mm on the other shot, and I don't want people putting these shots up against those ones and trying to draw some unsubstantiatable conclusion. So leave the other thread where it is, and compare these as a standalone entity please.

http://dvxuser.com/barry/7Dvs7D-still.jpg



http://dvxuser.com/barry/7Dvs7D-video.jpg


I agree that the still shot is indeed significantly sharper here, meaning that diffraction (and perhaps lens) did play an issue in the other test's overall stills sharpness. However, I believe the difference between the still and video is at least as great, if not greater, in this comparison, which leads me to conclude that the lens and diffraction did not hamper the video shot from the other comparison.

SuperApe
10-15-2009, 09:40 AM
Another test wow! To be honest since the video frame resolution is so much lower than the full still image there was never any way it would be the lens - maybe a little edge chromatic, maybe contrast, but very doubtful overall video frame sharpness.

This still *does* look sharper, but the last one was at sharpness 0 and the lighting looks more contrasty here.

Unless you personally have doubts, please don't spend more time re-proving your point, head on to the new test vs GH1!

MovieSwede
10-15-2009, 09:41 AM
Thanks for the test Barry.

Any idea why they cant make the videofootage look more like stillfootage?

Is it lack of processing power? Or the codec?

7 frames in 1 seconds should be easier then 24 frames in 1 second.

Everts
10-15-2009, 09:46 AM
Another test wow! To be honest since the video frame resolution is so much lower than the full still image there was never any way it would be the lens - maybe a little edge chromatic, maybe contrast, but very doubtful overall sharpness.

This still *does* look sharper, but the last one was at sharpness 0 and the lighting loks more contrasty here.

Unless you personally have doubts, please don't spend more time re-proving your point, head on to the new test vs GH1!


\
second that !

still vs frame grab ???

Martti Ekstrand
10-15-2009, 09:48 AM
Try importing that still into After Effects in original quality, scale it down in a 1080p comp 10 seconds long and render it out mimicking as close as possible the h.264 settings of the 7D. Just to check if that's a complicit in the reduction of sharpness.

patssle
10-15-2009, 09:53 AM
Now that is a good sharp still photo. I still wonder about your last thread, because the still photo from that one looks blurry.

In video mode, did you do 720 60 or 1080 30 or 24?

Isaac_Brody
10-15-2009, 09:55 AM
Thanks Barry, I just have one small and pertinent complaint. We should judge this footage when it's in motion. If you send me a link I'll upload it later tonight on vimeo so people can see it full quality and raw.

ELabree
10-15-2009, 09:58 AM
Why does the video look like SD? Are both a crop of a larger image?

commanderspike
10-15-2009, 09:59 AM
Is it lack of processing power? Or the codec?

Try downscaling 24 18MP images to 1080p in Photoshop one after the other on your PC. See how long it takes. Even with dedicated hardware routines on chip, a portable electronics with a current generation CPU is never going to be able to find the time to do that 24 times a second! That much is obvious.

So yes, it's a lack of processing power AND the codec. That's why the image ends up the way it does. I've seen as much by doing the same test on the GH1 but on the GH1 colour also seems to suffer, not just detail.

In my opinion, this will be the main area of improvement coming in the future to high end VDSLR models, and there is a rumour that Digic V will be based around a video core and featured in the 1D Mark IV. But just to be clear, at the moment that's just speculative.

Barry_Green
10-15-2009, 10:01 AM
Any idea why they cant make the videofootage look more like stillfootage?

Is it lack of processing power? Or the codec?
Well, to me, that's what's fascinating in all this. I look at what we currently have, which isn't bad, but it's nowhere near the "hail mary full of grace" that folks had been whipped into a frenzy to believe. But then I look at what it could be, and ... man... it's soooo close...


7 frames in 1 seconds should be easier then 24 frames in 1 second.
Yes, but (and, as Pee Wee Herman said, everyone always has a big but): that's the big question. 7 frames in 1 second is much easier than 24 in 1 second, and that's why we can get the better quality in 7 frames. What's the holdup between 7 and 24? It's definitely hardware, as Canon has already said. It shouldn't be the codec, because we already have great AVC codecs that easily handle 1080/24p. So I put the suspects as being the way the chip is read (the binning-6-pixels technique) and the hardware downscaler.

You cannot make a single camera that's optimized for stills, and also for video (unless, of course, you take Red's approach, wherein the stills are the same resolution as the video, last I heard). So when you have a big disparity between still size and video size, one or the other is going to suffer. On the 7D, it's primarily a still camera, and the still performance is optimized, and the video performance is compromised. On the HCM40, which is primarily a video camera, the video performance is optimized and the still performance is compromised.

Canon said that they know what we want. They said that it's hardware that's the difference. And they said that they'd build it, and that it'd be more expensive.

If anything, this has got me jazzed up for what Canon, Scarlet, and (presumably) a GHX100 could deliver. The imaging is there, deep down, they just have to design the hardware to be optimized for video usage.

commanderspike
10-15-2009, 10:02 AM
We should judge this footage when it's in motion.

Indeed, even though one frame isn't too detailed - if your eyes see it 24 times a second in motion this gives the illusion of their being more detail than there actually is in just one still frame.

Tests like this are scientific, but the proof of the pudding is often in the eating. I am more looking forward to the cinema projection tests, of not still frames - but motion images.

Barry_Green
10-15-2009, 10:03 AM
Try importing that still into After Effects in original quality, scale it down in a 1080p comp 10 seconds long and render it out mimicking as close as possible the h.264 settings of the 7D. Just to check if that's a complicit in the reduction of sharpness.
Well, see, that was part of what I wanted to do with the first test, which is why I shot it in motion. But I didn't accurately compensate for the shutter speed's effect on the frame rate, so it wasn't a direct simulation.

I'd like to do exactly what you suggest, but I'd have to find a way to get a motor-drive still sequence that is comparable to the full-speed motion sequence.

I guess I could get there if I used a super-high shutter speed... but then it wouldn't be a fair motion test...

Barry_Green
10-15-2009, 10:05 AM
Thanks Barry, I just have one small and pertinent complaint. We should judge this footage when it's in motion. If you send me a link I'll upload it later tonight on vimeo so people can see it full quality and raw.
Can't please everyone. :)

Can't judge it in motion, because the still was a still. And the motion would be wrong, because I had to go to 1/400 shutter to get the iris up to f/5.6.

There is no way to do what everyone wants, because the equipment is not capable of it. Unless, of course, I do it as a night shoot, and bring out the HMIs and light up the whole street, but light it to a controllable level so that I can get 1/50 shutter and f/5.6... and while I could do that, I'm not gonna. :)

morgan_moore
10-15-2009, 10:05 AM
What is this proving ?

1080 is about 10 times smaller than 18mp (I think)

cut the res by 10 and it looks 10X worse

surely we know that

Or do you htink it looks 20X worse - which would be interesting or what?

It would be great to see the 7 against a 5d, EX1 and D3s

my casual observations is that it matches the EX1, is not as good as the 5d and is potentially equalled by a good 720p - we dont know if the nikon is good of course


-----

to me the way forward would (if I were a DSLR designer) try and get normal still picture shoot rate up to 24/25 and port the data out onto an external recorder, one with XLR inputs

with the D3 getting 11FPS they only need to double it, which in terms of processor speed is not that hard given development time elapsed since the D3 (not S) launch

S

Barry_Green
10-15-2009, 10:07 AM
Try downscaling 24 18MP images to 1080p in Photoshop one after the other on your PC. See how long it takes. Even with dedicated hardware routines on chip, a portable electronics with a current generation CPU is never going to be able to find the time to do that 24 times a second! That much is obvious.
And certainly not at $1700! Remember, the price tag is going to play a part in this. The more you spend, the more power they can build into the hardware.

Barry_Green
10-15-2009, 10:08 AM
Why does the video look like SD? Are both a crop of a larger image?
Yes, I should have specified that -- both are pixel-for-pixel extractions from 1920x1080 frames.

The still image was 5100-something wide, I cropped and resized it to 1920x1080, and then took these pixel-for-pixel extractions.

commanderspike
10-15-2009, 10:08 AM
So what is the solution?

As far as I understand it there is also another limitation - the sensor. Generates too much heat to output such a high resolution image.

Downsampling is such a waste of CPU time. The best solution is not to have a sensor designed for stills in the first place!

Ian-T
10-15-2009, 10:08 AM
Indeed, even though one frame isn't too detailed - if your eyes see it 24 times a second in motion this gives the illusion of their being more detail than there actually is in just one still frame.
.Exactly, which is one of the main reasons why the "Titanic" director wants the ability to shoot 48p for 24p playback.

I don't see why folks are so surprised. The still capture from any camcorder/DSLR hybrid is ALWAYS much sharper than the video. I thought this is common knowledge.

Barry_Green
10-15-2009, 10:09 AM
Now that is a good sharp still photo. I still wonder about your last thread, because the still photo from that one looks blurry.
Agreed, which is attributable to f/20 on the iris, vs. f/5.6 here, and the Canon kit lens vs. the Zeiss ZF.

However, I still don't think those differences made any appreciable difference in the relative difference between still and video mode.


In video mode, did you do 720 60 or 1080 30 or 24?
Only always ever 1080/24p.

Uwe Lansing
10-15-2009, 10:11 AM
This framegrab looks a lot more pleasing. All that in motion and nobody would say it is not sharp enough. But if so, a little sharpening in post will do the rest....

http://img362.imageshack.us/img362/5117/7dvs7dvideosharpen.jpg (http://img362.imageshack.us/i/7dvs7dvideosharpen.jpg/)

dadoboy
10-15-2009, 10:12 AM
On later thought, I didn't think that diffraction had much bearing on your previous comparison, because your 7D still taken on the right had ample detail.

But it's good to see these new stills with the lens changes. Thanks.

commanderspike
10-15-2009, 10:14 AM
Reacts very well to sharpening.

Can you do the same with this GH1 frame grab please capice, and see how it reacts?

http://www.brawnf1blog.com/GH1/movie720.JPG

******

For reference here is the downscaled still photograph of the same scene with the same lens, settings, etc.

http://www.brawnf1blog.com/GH1/still720.JPG

Barry_Green
10-15-2009, 10:22 AM
What is this proving ?
It's proving that it's not the lens, it's not the aperture, it's not anything anyone can control. The limiting factor that makes the 7D soft is the hardware in the camera.


1080 is about 10 times smaller than 18mp (I think)

cut the res by 10 and it looks 10X worse

surely we know that
But that's not what's being done. The 18 megapixels is being downscaled to 1080p in two ways -- either from the internal hardware scaler, or by photoshop after the fact. It's showing that there is tremendous potential for image quality improvements.


It would be great to see the 7 against a 5d, EX1 and D3s
Someone else would have to do that. I don't have a 5D, EX1 or D3s.


my casual observations is that it matches the EX1, is not as good as the 5d and is potentially equalled by a good 720p - we dont know if the nikon is good of course
My guess is that the EX1 would handily spank the 7D and 5D. the 7D and 5D are a good match for a native 720p camera, but not on par with a 1080p camera (even a 1/4" one, much less a 1/2" one).


to me the way forward would (if I were a DSLR designer) try and get normal still picture shoot rate up to 24/25 and port the data out onto an external recorder, one with XLR inputs
If someone wants a hybrid still and video camera, maybe. If all they need is a video cam, the potential is there, we just need better hardware reading/downscaling of the chip. That's why I really hope that when/if Canon does an XLH2 with this (or comparable) sensor, or when Panasonic does a GHX100, they just *forget* about the stills capability, and tune everything for optimal motion imaging.

If nothing else, running these tests has re-whet my appetite for a Scarlet.

Barry_Green
10-15-2009, 10:24 AM
On later thought, I didn't think that diffraction had much bearing on your previous comparison, because your 7D still taken on the right had ample detail.

But it's good to see these new stills with the lens changes. Thanks.
I think diffraction was not in any way limiting the video performance, but it may have been holding back the still performance. So with this latest comparison, there ain't nothin' holdin' nothin' back.

sblfilms
10-15-2009, 10:30 AM
I definitely think it will be nice when Canon finally releases a camera with enough horsepower to scale the full resolution image down to 1080p instead of the binning or line skipping that is robbing detail the sensor is fully capable of capturing.

I actually would LOVE for somebody to use a 3840x2160 resolution sensor as it scales to 1080p with relative ease and gives the best bet for reducing artifacts that often get introduced when scaling (up or down)

morgan_moore
10-15-2009, 10:31 AM
Barry - I think I understand - you have downsacled the (photo) image in PS and then cropped it

therefore showing what 1080 should or could look like

If that is the case then indeed you are showing what a POS the 7d video mode is

As and EX1 owner it think it is similar and not as good as the the 5d, (which I did own)

So given no RED budget what is the camera to have right now - that is the question?

We need to understand the D3s and see the 5d against it


---

A further thought on the 7d compression technology,

There is a possibility that basically the Data per image is restricted, hence good results with 50 at 2.8 which due to blur restrict data

Im thinking that the sweet spot of the cam is longer lenses and less DOF, to keep that data rate down

BTW - thanks for your efforts

S

morgan_moore
10-15-2009, 10:33 AM
I definitely think it will be nice when Canon finally releases a camera with enough horsepower to scale the full resolution image down to 1080p



I dont see the point of any downscaling in camera just give me 24/25 raw frames per second (onto a badass external recorder)

This would seem to put less stress on the cameras guts and give 5k res

Let a computer do the downscaling later

S

Uwe Lansing
10-15-2009, 10:37 AM
...
Can you do the same with this GH1 frame grab please capice, and see how it reacts?...


This framegrab looks pretty good to my eyes. I wouldnīt do any sharpening on it. But ok, here with the same settings as the other one...

http://img355.imageshack.us/img355/9048/movie720sharpen.jpg (http://img355.imageshack.us/i/movie720sharpen.jpg/)

commanderspike
10-15-2009, 10:42 AM
Thanks. I am now a little confused as to how my GH1's 17mbps AVCHD seems to out resolve the 7D in video mode, because Barry's frame grab is awfully blurry. Or maybe putting the 7D frame grab next to an insanely detailed still frame makes it look a lot worse to-the-eye than it actually is?

JoeC
10-15-2009, 10:43 AM
It would seem, and make sense that the problem is in the pixel binning rather than the compression. In order to use the full size of the chip, it looks like they've designed a group of pixels to act as one giant pixel - thus a loss of sharpness. This image is then compressed. The better solution perhaps would be to not group the pixels, let each pixel act as it does in the still mode - then in the process of compression, the image would be resized and compressed. This would probably require greater processing power than the camera can do however. If they could come out with a raw mode or the CF card adapter that feeds a computer - perhaps this would work.


-Joe

Barry_Green
10-15-2009, 10:44 AM
I dont see the point of any downscaling in camera just give me 24/25 raw frames per second (onto a badass external recorder)

This would seem to put less stress on the cameras guts and give 5k res

Let a computer do the downscaling later

S
For that, you want a Red. That's exactly Red's approach.

commanderspike
10-15-2009, 10:45 AM
And it could also be down to the fact that Barry presents his video frame at 1:1, whilst I present the full GH1 frame as downscaled from 1080p to 720p for the forum. At 1:1 it'd look a lot worse versus a GH1 RAW photo at 1:1. So bear that in mind when viewing the above :)

Remember also that on a computer, we sit a lot closer to the screen. Project the image onto a cinema screen and sit a long way back from it, and have the image moving, and it'll look a lot better. In my opinion I don't think this kind of testing is very relevant to the real world.

ydgmdlu
10-15-2009, 10:49 AM
with the D3 getting 11FPS they only need to double it, which in terms of processor speed is not that hard given development time elapsed since the D3 (not S) launch
Please note that the 11 fps is using the mechanical shutter. So I feel that the mechanical shutter is a major limiting factor to getting what you want. The entire sensor can be scanned much faster than that, as the video modes prove. The mechanical shutter is there specifically to improve image quality while shooting stills. This cannot be repeated or emphasized enough.

JamesBenet
10-15-2009, 10:50 AM
Great Test, thanks for not prolonging the agony by just giving all the info since the start. The other test made me close to schizo! Horrible video quality compared to stills. If the red down-sample from Scarlet looks anything like that 18MP still down-res then Jim and Co. have nothing to worry about for at least 2 years.

Now how about a test just like this but of a closer subject "not charts" like people or some stuffed dolls? I feel that the closer stuff is to these cameras the better they look according to the videos.

Now most of us have less than perfect vision and if you see foliage you are closer to the video image than the 18MP image in your brain, maybe thatīs why the image is still so appealing, looks closer to our flawed vision!

patssle
10-15-2009, 10:52 AM
Remember also that on a computer, we sit a lot closer to the screen. Project the image onto a cinema screen and sit a long way back from it, and have the image moving, and it'll look a lot better. In my opinion I don't think this kind of testing is very relevant to the real world.

And also, how often does one deliver on Blu-Ray or project onto a screen? Most of the time it's downsampled for the web or DVD, which makes it look great sharpness wise.

ydgmdlu
10-15-2009, 10:52 AM
If they could come out with a raw mode or the CF card adapter that feeds a computer - perhaps this would work.
It wouldn't work because the image is scaled and compressed before it's recorded to the memory card, so sticking a RAW adapter or module does nothing to circumvent the scaling and compression. The video is derived from the live view stream. Blame the reduced quality on that as much as anything else.

morgan_moore
10-15-2009, 10:56 AM
Please note that the 11 fps is using the mechanical shutter. So I feel that the mechanical shutter is a major limiting factor to getting what you want. The entire sensor can be scanned much faster than that, as the video modes prove. The mechanical shutter is there specifically to improve image quality while shooting stills. This cannot be repeated or emphasized enough.

Are you saying that 25FPS should be even easier than waiting for computer development because the mirror action is the limiting factor and of course could be locked for motion recording

or are you saying that the mirror needs to come down as part of the process of recording a frame therefore requiring a significant upgrade to the mechanics rendering my dream cam unlikely because mechanical development is slower than processor speed development?

I know that my efforts as using my hasselblad digiback as a pin hole camera where futile because the camera needed to see black after the exposure to create a 'black reference'



S

commanderspike
10-15-2009, 11:00 AM
Rolling shutter on a 18MP sensor with a raw read out would be hellish. Correct me if I'm wrong but the higher the data output of the sensor the slower it is to process it, top to bottom.

As far as I understand it the mechanical shutter is kept for stills to avoid exactly that issue.

Barry_Green
10-15-2009, 11:03 AM
The entire sensor can be scanned much faster than that, as the video modes prove.
Ah, but that's the question -- CAN the entire sensor be scanned that much faster? Red has a video-oriented large CMOS sensor, and they can't scan the entire sensor at high frame rates. The most they can get is 30fps out of it in full-res mode.

Yet the 7D can do 720/60p, with the full field of view.

Is the 7D doing something that the 10x-more-expensive Red cannot do? I would think not. So therefore it makes me think that the Canon is definitely not scanning the entire sensor. When Red goes into 2K or 3K mode, it's scanning a portion of the sensor, and can therefore scan it more quickly (which is why they get up to 120fps in 2K mode; they're scanning 1/4 of the sensor, and can do so 4x as fast). So the 7D is almost certainly not scanning the entire sensor. Some speculated that it was actually skipping pixels, Canon says they're binning them (which means reading multiple pixels as part of one big larger pixel). Either way, they're certainly not scanning 18mpix at 60fps...

Barry_Green
10-15-2009, 11:05 AM
Now how about a test just like this but of a closer subject "not charts" like people or some stuffed dolls? I feel that the closer stuff is to these cameras the better they look according to the videos.
Sure it will, and largely that's what we've been seeing (such as Dublin's People, etc). Even an SD DVX will look great on a closeup. These shots were tests to find out what the real resolving power would be.

We already know that the 7D, 5D, and GH1 excel at closeup shots. Not really any reason to test that.

(unless, perhaps, testing a closer shot on the 7D vs. on the HMC40/Letus, which would let us know if there's anything about the DSLR that's exceeding what the video camera can do, other than sensitivity)... hmmm...

commanderspike
10-15-2009, 11:05 AM
Any theories as to why Canon haven't released a low megapixel full frame 35mm film sized CMOS video imager yet, since they seem to have the technology - why not just lower the pixel count to 2k native and make the photosites larger?

Barry_Green
10-15-2009, 11:06 AM
Most of the time it's downsampled for the web or DVD, which makes it look great sharpness wise.
Exactly, and that's what Canon designed these cameras for. Photojournalists who were also going to have to post footage for the web. Well, that's what they said they specifically designed the 5D for; the 7D is getting lumped in there by me because it appears to use the same caliber of hardware.

And guys, don't forget this -- it's $1700. I've paid more than that for a mattebox. My tripod cost more than that. I've bought individual light fixtures for more than that.

commanderspike
10-15-2009, 11:08 AM
Also, any ideas why Canon haven't jumped on the GH1 style full-time live view bandwagon yet? Because from a technology point of view, I could almost disassemble my Mark II and take out the mirror box and view finder with a damned screw driver. They have the GH1 style mechanical shutter, they have the sensor, they have the live view. Get rid of the mirrorbox and put a EVF in there. It seems so straight forward, doesn't it?

KeithAndrews.TV
10-15-2009, 11:09 AM
Ok, so it's the hardware. What is the possibility of funding someone like Tramm Hudson and Barry for example, to rewrite the firmware, like in Magic Lantern, so that the camera can be MORE optimized for video as opposed to still? Notice I said more optimized, not completely optimized for video. This might entail finesing the video codec and compression rates as well as adjusting the way the sensor downconverts the image.

morgan_moore
10-15-2009, 11:10 AM
Rolling shutter on a 18MP sensor with a raw read out would be hellish. Correct me if I'm wrong but the higher the data output of the sensor the slower it is to process it, top to bottom.

As far as I understand it the mechanical shutter is kept for stills to avoid exactly that issue.

No comprende ?

there is no jello in still photogrraphy

or is that actually a still frrame taken at readout speed with the shutter , restricting the capture to the shutter time ?

ie you are saying that my just grab the raw concept would require a mechanical shutter and therefore is a no-go

of course one wonders if a blinking LCD or OLEd in front of the chip could 'create' a shutter or even a cinema shutter built into the lens or put between MF lenses and the cam body

S

ydgmdlu
10-15-2009, 11:10 AM
Are you saying that 25FPS should be even easier than waiting for computer development because the mirror action is the limiting factor and of course could be locked for motion recording

or are you saying that the mirror needs to come down as part of the process of recording a frame therefore requiring a significant upgrade to the mechanics rendering my dream cam unlikely because mechanical development is slower than processor speed development?

I know that my efforts as using my hasselblad digiback as a pin hole camera where futile because the camera needed to see black after the exposure to create a 'black reference'



S
It's not the mirror. It's the shutter, which is a separate mechanism. The mechanical shutter on a DSLR is very much like the shutter on a film camera. Here's a simple test to see what I mean: Take your 7D and activate live view. Now put your finger under the mirror. Deactivate live view. Your finger should prevent the mirror from moving back to its default position. So you should have a clear view of the sensor. Now press the shutter button and watch what happens with the sensor.

commanderspike
10-15-2009, 11:11 AM
Ok, so it's the hardware. What is the possibility of funding someone like Tramm Hudson and Barry for example, to rewrite the firmware, like in Magic Lantern, so that the camera can be MORE optimized for video as opposed to still? Notice I said more optimized, not completely optimized for video. This might entail finesing the video codec and compression rates as well as adjusting the way the sensor downconverts the image.

I don't think two individuals can do a better job of Canon on optimising the image for video. Feature set, as we've seen with Magic Lantern on the 5D, maybe. But the downscaling is an altogether higher level issue and very much hardware (not firmware) restricted - and hence also even more price restricted.

commanderspike
10-15-2009, 11:13 AM
No comprende ?

there is no jello in still photogrraphy

or is that actually a still frrame taken at readout speed with the shutter , restricting the capture to the shutter time ?

ie you are saying that my just grab the raw concept would require a mechanical shutter and therefore is a no-go

of course one wonders if a blinking LCD or OLEd in fron of the chip could 'create' a shutter

S

That's my point. There is no jello in still photography, because as far as my knowledge goes, the mechanical shutter prevents this.

Video needs an electronic shutter, and this along with read-out speed defines the severity of jello. Correct?

KeithAndrews.TV
10-15-2009, 11:13 AM
I don't think two individuals can do a better job of Canon on optimising the image for video. Feature set, as we've seen with Magic Lantern on the 5D, maybe. But the downscaling is an altogether higher level issue and very much hardware (not firmware) restricted - and hence also even more price restricted.

Well, you never know. We already know its a stills camera optimized for stills that just happens to have video, so maybe we can tweak it a bit and make the video mode that much better.

ydgmdlu
10-15-2009, 11:14 AM
Also, any ideas why Canon haven't jumped on the GH1 style full-time live view bandwagon yet? Because from a technology point of view, I could almost disassemble my Mark II and take out the mirror box and view finder with a damned screw driver. They have the GH1 style mechanical shutter, they have the sensor, they have the live view. Get rid of the mirrorbox and put a EVF in there. It seems so straight forward, doesn't it?
It's because photographers prefer the optical viewfinder. Just ask TimurCivan. When Canon makes a dedicated video product, it will have EVF instead of OVF.

The GH1 was not intended for the same market as Nikon and Canon's DSLRs, which are more for prosumers and professionals. The GH1 was intended for amateurs and casual photographers interested in DSLRs but unwilling to make the full leap from compact P&S digicams.

shoqman
10-15-2009, 11:16 AM
This is a fascinating test. After my comments to yesterday's tests, I have shot more stuff and concluded that shooting with sharpness at 0 or 1 is best, and then more sharpening in post when needed. The footage looks awesome sharpened in post, and you really only need to do it to shots that are deep focus. It's not a problem with any other kind of shot.

That being said, it does appear the GH1 has better scaling hardware. But frankly, for me, the use of random non-standard MTS wrapper and pulldown requiring transcoding and pulldown removal before I can do anything with the footage, plus the poor codec quality means it's totally useless to me.

To be able to pop a compact flash card into my monitor and drop it immediately into any timeline is huge. I can live with sharpening a handful of wide, deep focus shots.

But I'm not ignoring the fact that the 7D appears to be significantly worse at these kinds of shots than other cameras. I'm also not surprised by it, or concerned by it.

I bought this camera for the kinds of shots that I have previously been adapting those other cameras to do, and it KILLS them for those kinds of shots.

morgan_moore
10-15-2009, 11:16 AM
It's not the mirror. It's the shutter, which is a separate mechanism. The mechanical shutter on a DSLR is very much like the shutter on a film camera. Here's a simple test to see what I mean: Take your 7D and activate live view. Now put your finger under the mirror. Deactivate live view. Your finger should prevent the mirror from moving back to its default position. So you should have a clear view of the sensor. Now press the shutter button and watch what happens with the sensor.

Sorry my unclear language, I know the difference between shutter and mirror !

You are saying that a shutter is required to record still style images

(and therefore with the mechanical in appropriatness of DSLR shutters it could not happen without significant mechanical development - which is slow unlike electronic development)

S

JamesBenet
10-15-2009, 11:16 AM
Also, any ideas why Canon haven't jumped on the GH1 style full-time live view bandwagon yet? Because from a technology point of view, I could almost disassemble my Mark II and take out the mirror box and view finder with a damned screw driver. They have the GH1 style mechanical shutter, they have the sensor, they have the live view. Get rid of the mirrorbox and put a EVF in there. It seems so straight forward, doesn't it?

Well because optical viewfinders are way faster and clearer. Even the GH1 suffers in low light and slows down like hell in low light. Light reaches your eye almost in real time by a few nanoseconds due to the lens and the mirror/prism so an electronic viewfinder with a 1/8th of a second lag is immensely slower .

Its also a photo oriented camera and any photo centric users who saw an electronic laggy image would go ape if that ever happened. :Drogar-Shock(DBG):

Now tech is getting there, see the Epson post in dpreview news! I still think its not ready at that res or refresh rate! Maybe in 3 years!

ydgmdlu
10-15-2009, 11:17 AM
That's my point. There is no jello in still photography, because as far as my knowledge goes, the mechanical shutter prevents this.

Video needs an electronic shutter, and this along with read-out speed defines the severity of jello. Correct?
Exactly.

ydgmdlu
10-15-2009, 11:21 AM
Any theories as to why Canon haven't released a low megapixel full frame 35mm film sized CMOS video imager yet, since they seem to have the technology - why not just lower the pixel count to 2k native and make the photosites larger?
There's no market for it yet. Canon's not in the digital cinema business, yet. Until they get in that business, I see no reason to assume that they have a serious interest. Digital cinema isn't exactly big business; RED's not in it for the big bucks. And if Canon were to create the sensor that you want, it wouldn't be sold for much less than what RED's offering.

Also, there's no full-frame digital cinema camera on the market. The 5D Mark II and the Nikon D3s are the only video-capable cameras with full-frame sensors. Cinematographers are not used to full frame and won't demand it; cinematography equipment isn't designed for full frame either.

Jean Dantes
10-15-2009, 11:23 AM
This is at sharpness +3? Would have expected it to be a little better...

Rory_B
10-15-2009, 11:34 AM
I don't see why the 7D and the Gh1 need to be mutually exclusive. Considering their price point you could get both and use each to their advantages and inter-cut them quite easily. The 5D,red, GH1 video showed that.

And yes I know that was the 5D not the 7D in the view, but since the 7D and GH1 both shoot 24p you've got no issues editing on the same timeline after encoding to a more edit friendly format.

Buck Forester
10-15-2009, 11:41 AM
I apologize for not having the time to read the entire threads so disregard my comments if necessary, ha!

First of all my questions/comments are not defensive nor biased to any manufacturer. They are generic, based on concepts. I don't know much about the 7D and I'm not in the market for one, nor am I a rabid Canon fanatic, ha!

Obviously a still from a video will not compare in any way to a still from a still camera. I think the 7D shoots what, 18 MP? Even if you downsize and crop an 18 MP still to 1920x1080, you have a LOT more information in the image than shooting a 1920x1080 video. And of course a video mode shot is cramming 24 images per second at 1920x1080, so when you take a sampe random still from one of those images cranking through, it's not going to look like a still shot, even a still shot at the same 1920x1080 resolution. Not even close. That's just the physics of still vs video, at least where we are today. That's why a time lapse shot from multiple digital stills at full resolution is so stunning compared to in-camera time lapse when blown up big. Maybe the RED stuff will narrow this gap? I dunno, I hope so, but the gap will certainly be narrowed eventually.

Billy Barber
10-15-2009, 11:41 AM
Barry- hey Bill here. Listen man UPS just called and said you got the wrong 7D delivery. Apparently you got the one I was sending back to Canon after I dropped it on the sidewalk outside my house.

Hope this doesn't screw anything up.

Cheers!

Kellar42
10-15-2009, 11:51 AM
This is all very interesting.

A couple of questions and thoughts.

On that still you fellows sharpened in post...what kind of video filters would one apply in Final Cut Pro say to get a decent sharpening? My experimentation with the basic filter seems to just screw things up.

Also, I think the reason Canon and Panasonic downscale things in camera is that most people buying cameras at this price point don't have the capability to deal with 2-5k raw video. Heck, many people on this forum are still coping with getting their computers up to speed to handle 1080p...imagine your consumer or aficionado or even semi-semi-pro picking up something at Best Buy for less than two grand and trying to get his movies onto a DVD at home?! The only cameras really shooting Raw, like the Red aren't something anyone is going to stumble on by accident.

And just an observation, but some frame grabs of a recent Opera I shot with the GH1 have been good enough for facebook, websites, and online press as actual photos. They aren't as hot at as the stills, obviously, but are significantly better than grabs from say an HV30 and what I'm seeing here on the 7D.

Michael Olsen
10-15-2009, 11:56 AM
Also, I think the reason Canon and Panasonic downscale things in camera is that most people buying cameras at this price point don't have the capability to deal with 2-5k raw video.

This is part of the reason, I'm sure. But I think a much greater reason is that, in order to put out a 2-5K RAW image a camera needs to have very specific hardware capable of doing that. In this instance, it doesn't seem that the sensor can scan quickly enough to record a 5K RAW image. Nor can the CF card record fast enough, nor hold more than a few seconds of footage. Compression would always be an option, but the DIGIC IVs aren't going that great a job with the 1080p H264, not to mention compressing 5K RAW.

Unfortunately, Canon still refuses to give us all something great for nothing. We get pretty close to something for nothing with the 7D, though.

ydgmdlu
10-15-2009, 11:58 AM
Unfortunately, Canon still refuses to give us all something great for nothing.
Not even RED will do that!

morgan_moore
10-15-2009, 12:01 PM
Even if you downsize and crop an 18 MP still to 1920x1080, you have a LOT more information in the image than shooting a 1920x1080 video.

You dont because you just threw that information away by downsizing

Barry has correctly demoed what the 7d IS doing and what it could/should be doing

of course it is unlikeily that any other sub $20k camera is doing that either

which is what my telly mate and I have been arguing about since I got my 5d, he reckons his old Beta Cam ($50k?) is better even though it is SD !

this was indeed clarified mid thread

S

Buck Forester
10-15-2009, 01:05 PM
You dont because you just threw that information away by downsizing

But morgan, when you start off with lots of information and downsize, you have a lot more non-essential information to toss. So the resulting information is much better. This is a basic principle, no?

What makes a better print? An 18 MP file editing and downsized to make a 1920x1080 print, or a shot that was 1920x1080 to begin with?

That's why photographers shoot RAW files at full size and convert to TIFF (lossless), rather than starting with an in-camera jpeg at the size they plan to print. You have a lot more information and when/if you compress (discard) you have more 'useless' information to toss than if you shot it at 1920x1080.

Eddy Robinson
10-15-2009, 01:10 PM
Yeah, when you downsize in ost you can do content-sensitive image analysis, that's not as easy in realtime.

Ian-T
10-15-2009, 01:42 PM
Supersampling!!!

JoeC
10-15-2009, 02:01 PM
SO let me get this straight -

If I bought one of these....

http://www.joecrouch.com/24canon.jpg

and set each camera to shoot stills at 8 fps... I would have a killer 24P camera system.

-Joe

morgan_moore
10-15-2009, 02:14 PM
For five second clips

S

morgan_moore
10-15-2009, 02:17 PM
Seriously - Back to the topic

This is a concerning image

is there any side to side 5d material??

Understanding little of compression do we feel you actually get better resolution for less complex scenes ?

S

commanderspike
10-15-2009, 02:51 PM
Seriously - Back to the topic

This is a concerning image

is there any side to side 5d material??

Understanding little of compression do we feel you actually get better resolution for less complex scenes ?

S

Excuse the freaky doll *cough*

5D movie frame (1080p at 1:1 crop)

http://www.brawnf1blog.com/GH1/doll5dmovie.jpg

GH1 movie frame (1080p at 1:1 crop)
http://www.brawnf1blog.com/GH1/dollgh1movie.jpg

Compare it to the 5D still frame (22MP at 1:1) (http://www.brawnf1blog.com/GH1/doll5d.jpg) and the GH1 still frame (12MP at 1:1) (http://www.brawnf1blog.com/GH1/dollgh1.jpg)

(Images a little too big for embedding)

Both F8 with Zeiss lenses. ISO 640, manual WB. GH1: smooth. 5D: faithful.

A few observations: both have artefacts, both a little soft (especially compared to the stills) but the 5D has less nastiness. Macroblocking and compression is extreme on the GH1, turning a lot of the fine detail to mud. I have no idea how the 7D compares to the 5D in that respect. Probably gives a similar image to the 5D. Comparing a still from a camera as good as the 7D or 5D to it's video output is a little cruel.

But it's clear the 5D has a better image to use as a spring board for movies, at least compared to the GH1.

Remember that these may be viewed by us as video cameras, but it's not just the downscaling that matters - it's the still image at 5K as well. Just look at the difference in detail between the 5D's image at 1:1 and the GH1's. As well there might be a big difference - double the resolution and sensor size on the 5D is no small matter.

Now, as an aside. Viewing this material on a 50 inch screen is quite dramatic. Imagine blowing the compressed GH1 1:1 footage up to 3 times the size of your computer screen. The differences between the 5D and GH1 are exaggerated. Toggling between the 5D's video and the 5D's still produces barely any discernible difference, but toggling between the GH1's photograph and it's video of the same scene, well, let's just say the video comes off pretty wretchedly at 50 inch, to say the least.

Now somebody lend me a 7D so I can test that as well :)

I think with the 7D, you don't need to regret your purchase because Barry did a 40 page test, milking every post for all it's worth. See the footage of the 7D for yourself on a big screen and then decide if it's worth the money.

YOU are the boss.

Ben_B
10-15-2009, 03:11 PM
Commanderspike you're breaking a forum rule by pointing the GH1 at something red. Stop it.

commanderspike
10-15-2009, 03:17 PM
The forum rules clearly stated that only green objects are allowed in camera comparisons (in tribute to Barry Green) but because I'm such an argumentative sod, I'd thought I'd break them anyway :)

Barry_Green
10-15-2009, 03:24 PM
Also, any ideas why Canon haven't jumped on the GH1 style full-time live view bandwagon yet?
Liveview on the 7D has a limited amount of time it can run, doesn't it? On the GH1 it's unlimited. I don't know why that is, but I believe there's a difference there.


Get rid of the mirrorbox and put a EVF in there. It seems so straight forward, doesn't it?
For a video camera, certainly. But for a still camera, no way -- and the 7D is even more a still camera than the GH1 is. People are clamoring for an optical viewfinder on the Red One even. I don't expect Canon to ditch the optical any time soon.

The GH1 got great compactness by getting rid of the mirror, but... no, side by side, the optical VF just slays any current electronic VF...

Barry_Green
10-15-2009, 03:25 PM
What is the possibility of funding someone like Tramm Hudson and Barry for example, to rewrite the firmware, like in Magic Lantern, so that the camera can be MORE optimized for video as opposed to still?
Wouldn't happen, and no real need to; in the six months it'd take us to develop something, Canon will introduce a new video-oriented product, and so will Panasonic. I'm pretty sure this situation we're in now is a temporary phase.

Ben_B
10-15-2009, 03:26 PM
The GH1 got great compactness by getting rid of the mirror, but... no, side by side, the optical VF just slays any current electronic VF...

That said, I really miss my Gh1's focus assist when shooting on a traditional SLR and trying to focus on distant objects.

Barry_Green
10-15-2009, 03:28 PM
There is no jello in still photography, because as far as my knowledge goes, the mechanical shutter prevents this.
Correct.


Video needs an electronic shutter, and this along with read-out speed defines the severity of jello. Correct?
Jello happens because of the way the sensor scans when in video mode; it progressively scans down the chip, instead of scanning the entire chip at once. It uses an "electronic shutter" in the sense that it turns a row of the chip sensitive, then it turns that row off and moves on to the next row, etc. And it takes around 1/24 to 1/60 of a second for the chip to scan all the way down, and that's a long time, plenty of time for the bottom of the image to have moved in comparison to the top of the image.

Whereas in still mode, it reads all at once, and then offloads all at once. But it's only really capable of doing that because of the very slow frame rates, as I understand -- you couldn't do that in video mode because there's not enough time to offload all the chip before you need it ready for the next exposure. Obviously, it's possible that technological advancement will overcome that difficulty, but as it stands now, it reads one line while offloading the prior line (by way of example, not as stated fact), so it's reading and writing as it goes, so there's no big "offload" stage.

Barry_Green
10-15-2009, 03:29 PM
The GH1 was not intended for the same market as Nikon and Canon's DSLRs, which are more for prosumers and professionals. The GH1 was intended for amateurs and casual photographers interested in DSLRs but unwilling to make the full leap from compact P&S digicams.
The GH1 was primarily about making an SLR-style camera as small as possible, from what I've read. Well, not true -- the G1 was about that, the GH1 was about adding video capabilities to the existing G1.

The 7D is a traditional DSLR in all ways, plus video. The G1/GH1 were new form factors introducing a new format, all about compactness etc. So no, not really directly competing for the same market.

Ben_B
10-15-2009, 03:30 PM
If you have questions about Jello, watch this, and ignore the insane narrator:
http://www.youtube.com/watch?v=whyzpwqXixU

Barry_Green
10-15-2009, 03:33 PM
That being said, it does appear the GH1 has better scaling hardware. But frankly, for me, the use of random non-standard MTS wrapper and pulldown requiring transcoding and pulldown removal before I can do anything with the footage, plus the poor codec quality means it's totally useless to me.
This is the kind of thing people should be comparing them on, in my opinion. They both can take great footage, they both can take less-than-great footage, they both have their problems, ad nauseum. But there are certain factors that are real advantages, one way or the other. The post workflow of the 7D is just plain superior to the GH1, no two ways about it. The GH1 has the ergonomics and usability factor, but the post is way inferior (well, for us NTSC guys; it really isn't much of an issue for the PAL folks). If the post is a major problem for you, that right there is enough to decide on a 7D instead of a GH1.


To be able to pop a compact flash card into my monitor and drop it immediately into any timeline is huge. I can live with sharpening a handful of wide, deep focus shots.
Agreed, that's my favorite aspect of the 7D by far. I've grown used to this workflow with all my other cameras, the P2 and AVCHD cams all allow it, but the GH1 has that stupid pulldown-removal step that prevents it, and it *is* annoying. The instant editability and no-pulldown is a great feature.


I bought this camera for the kinds of shots that I have previously been adapting those other cameras to do, and it KILLS them for those kinds of shots.
I appreciate your angle on this, this is exactly how I feel people would serve themselves best by considering. No one product does everything, but pick a product that does what YOU need done.

Barry_Green
10-15-2009, 03:34 PM
Unfortunately, Canon still refuses to give us all something great for nothing. We get pretty close to something for nothing with the 7D, though.
You get something great for almost nothing, that's for sure. I still can't believe how cheap these things are. I bought both of 'em, for less than my HPX170 cost me.

commanderspike
10-15-2009, 03:50 PM
Great eye opener there Barry, especially on how the sensor works in video mode as opposed to stills mode. Makes perfect sense. It's all about speed in video mode, so a few tricks have to be done, a few liberties taken.

If the 5D ever got it's 24p patch, that would mean the sensor would be read off at 1/24th of a second rather than 1/30th? Which would make jello worse.

I am not sure about the 7D's live view not being able to stay on, I have not used one yet, but on the 5D I can keep my live view on for a long time if I completely disable power management. How long exactly I am yet to keep track of, but it doesn't seem to turn itself off.

What I meant regarding the mirrorbox is that the 5D is essentially a video core designed into a stills camera. It has a fast CMOS - doesn't need that for stills - that's video technology! It has a electronic shutter and a live view chip. All ready to go Canon.

commanderspike
10-15-2009, 03:52 PM
Agreed, that's my favorite aspect of the 7D by far. I've grown used to this workflow with all my other cameras, the P2 and AVCHD cams all allow it, but the GH1 has that stupid pulldown-removal step that prevents it, and it *is* annoying. The instant editability and no-pulldown is a great feature.

I echo the GH1 pulldown complaint, but with the 5D and 7D you still have to transcode the footage to Prores 422 / LT or XD CAM 35mps to get acceptable editing performance in FCP.

I cannot edit the raw footage on my Macbook Pro Core Duo 2ghz. Toooo slow.

The 5D & 7D are recording in a finishing format, not a recording format.

Duke M.
10-15-2009, 04:13 PM
It really bothered me that the colors were so different between the still picture and the video capture. The video looks muted by comparison. It has that desaturated look needed for color grading video.

I took both pictures, didn't touch the still image. On the video I did a simple gamma correction.

Red = .78
Green = .96
Blue = .78

The street sign on the still is clearer than the video. However, the leaves at the bottom are now molded and organic looking on the video, while on the still the leaves break up into pixelation. [Note again, I did not change, resave or do anything to the still image.]

To my eye now the video looks sharper, curvy and organic. But they both look a LOT closer, IMHO.

Still:

http://www.dvxuser.com/V6/attachment.php?attachmentid=13293&stc=1&d=1255648341

Video:
http://www.dvxuser.com/V6/attachment.php?attachmentid=13295&stc=1&d=1255648636

Rory_B
10-15-2009, 04:19 PM
Liveview on the 7D has a limited amount of time it can run, doesn't it? On the GH1 it's unlimited. I don't know why that is, but I believe there's a difference there.


Live view can stay on as long as you want. Just go into the settings and turn "auto power off" from 1min to "off"

Ian-T
10-15-2009, 04:25 PM
Excellent job Duke.

Rory_B
10-15-2009, 04:26 PM
I echo the GH1 pulldown complaint, but with the 5D and 7D you still have to transcode the footage to Prores 422 / LT or XD CAM 35mps to get acceptable editing performance in FCP.

I cannot edit the raw footage on my Macbook Pro Core Duo 2ghz. Toooo slow.

The 5D & 7D are recording in a finishing format, not a recording format.

All depends on how fast your machine is, my mac-pro tower is fast enough to edit h264.

When I do CC, I simply change the compressor settings to animation or something similar or do a media manager simply of the clips I have used and transcode those particular files. Saves a lot of time.

J. Odoms
10-15-2009, 06:03 PM
I'd personally like to see one of these comparative tests done with the 5D vs. 7D, and see who comes out victorious.

ydgmdlu
10-15-2009, 06:21 PM
5D is essentially a video core designed into a stills camera. It has a fast CMOS - doesn't need that for stills - that's video technology! It has a electronic shutter and a live view chip. All ready to go Canon.
Nope. I don't know in what way you could say that the 5D (or 7D) has a "video core" more so than a "still photo core." The hardware was designed for stills, not video; the video functionality was a hack. You yourself have said time and again that Canon's (allegedly) planning a real video-oriented chipset for the 1D Mark IV. The CMOS in the 5D is not fast enough for general video purposes; consumer handycams priced at well under $1000 can beat it there. However, it is just fast enough to be acceptable for live view, which was originally intended as an aide for still photography, not as a step towards building video cameras. You can hack other Canon DSLRs to enable video recording as well. (http://www.engadget.com/2009/01/25/hack-provides-liveview-canon-eos-dslrs-with-video-recording/) Big deal.

commanderspike
10-15-2009, 06:49 PM
Sorry ydgmdlu you misunderstood me there, my fault. Let me make it a bit clearer...

By video core I don't mean the CPU core. I don't mean core in that sense. I just mean the innards of the camera, i.e. sensor, shutter, processor, LCD. A quick look inside the mirror box of your 5D and you can see that a lot of what Canon need to develop for a next gen video camera is already there. Just imagine it without the housing and mirror box.

This post was a bit off topic. Apologies for that.

commanderspike
10-15-2009, 06:52 PM
I'd personally like to see one of these comparative tests done with the 5D vs. 7D, and see who comes out victorious.

I would also like to see this test. Apparently they are supposed to be quite closely matched, but I'd like to know for sure.

ydgmdlu
10-15-2009, 07:01 PM
I disagree with you there, and I understood you just fine. Perhaps the rudiments exist, but the hardware, which includes all of the things that you mentioned, not just the image processor, is just not ready for general video use. Formal and casual tests have shown this. You can't do a direct component transplant like you're suggesting. There's at least one more generation of video-oriented development before the hardware is truly ready.

I'll repeat that video is just a hack, which is why it isn't anywhere near as good as it could be on these cameras. (That's my way of getting back on topic...) You could theoretically hack any live view camera to do the same thing. (In my post above, I even provided a link that shows you how to do it on the old Canon 40D.) Does that mean that every live view camera ever made has innards suitable for general video work?

Barry_Green
10-15-2009, 07:57 PM
I think the 5D was designed to deliver footage to the web, as Canon said -- that's why it's 30.000 fps-only, and why the downscaling hardware isn't the greatest (although still pretty darn good). I think with the 7D, they fixed the obvious easy things (like the frame rate, upping the bitrate on the codec) but it's the same basic generation of hardware. As vdgmdlu says, I think we're one generation away from having the full enchilada.

ydgmdlu
10-15-2009, 08:17 PM
With these DSLRs, every new camera seems to be a game-changer. :)

Even the Nikon D3S brings new things to the table: autofocus in a full-frame video DSLR, groundbreaking low-light performance. Game-changer!

Barry_Green
10-15-2009, 08:20 PM
Still band-aids though. Give us professional audio, professional monitoring, full-res images, zebras, WFM... an HPX170 with a GH1 chip or an XLH2 with the 7D chip...

ydgmdlu
10-15-2009, 08:22 PM
My wallet weeps at the thought. :(

(Yet I'm ready to shell-out for a Scarlet.)

alexdias
10-15-2009, 08:41 PM
Still band-aids though. Give us professional audio, professional monitoring, full-res images, zebras, WFM... an HPX170 with a GH1 chip or an XLH2 with the 7D chip...

LOVE IT!

Will buy it as soon as available!

alexdias
10-15-2009, 08:44 PM
I hate to speculate and usually just comment on products that are at least released, hopefully available, but the Scarlet has a very good change of being that option. Even at a bigger price point which is expected.

sblfilms
10-15-2009, 09:21 PM
I didn't realize people were having to transcode to cut 7D footage. My Quad Core Mac eats those files up. I've only had one frame drop during playback in FCP and no beachballs....

David G. Smith
10-15-2009, 09:55 PM
Still band-aids though. Give us professional audio, professional monitoring, full-res images, zebras, WFM... an HPX170 with a GH1 chip or an XLH2 with the 7D chip...

Well, Isn't that what Red is up to?

I am personally not going to run out and get a 7D right now, but I have to admit that it's introduction has me very excited, and makes me very excited about what is just around the corner. I also thank you for taking the time to do these tests and sharing them with us. Your input really adds a very down and dirty objectivity to our evaluations of the the creative tools that are available to us. Thanks again.

As for the 7D, while yes, it has first generation DSLR-HD video processing hardware, uses a delivery codec for acquisition, has serious sound issues and has production limitations (12 minute recording, over heating, ect.), but for roughly $2000.00, it is still an amazing camera. With interchangeable lenses, a large sized sensor and what appears to be amazing low light performance, there are going to be some very, very impressive images, and hopefully, some very enjoyable films made with this camera.

With the experience of the use of this camera, and the power of business competition, we should also be expecting to see some more amazing cameras just down the pike. To all those who have bought this camera and who are going to buy this camera, I am not saying you have made a mistake. Hell no. Go out with this amazing camera and make some images that kick our asses. Use this amazing camera to make some serious cash and when something better comes along, add that to your tool kit also.

Ben_B
10-15-2009, 09:55 PM
Still band-aids though. Give us professional audio, professional monitoring, full-res images, zebras, WFM... an HPX170 with a GH1 chip or an XLH2 with the 7D chip...

http://img29.imageshack.us/img29/861/ghx100.jpg

Woo.

Chris_TC
10-16-2009, 02:24 AM
I didn't read the entire thread, so I'm sorry if this has been asked, but did you have in-camera sharpening turned on?

I seem to be seeing thick double edges all over the place that shouldn't be there if this were sharpened properly (0.5 pixels radius) in post.

http://dvxuser.com/barry/7Dvs7D-video.jpg

Barry_Green
10-16-2009, 07:09 AM
I didn't read the entire thread, so I'm sorry if this has been asked, but did you have in-camera sharpening turned on?

People, do you now see why it's impossible to satisfy everyone?

Chris, look at my other thread, you'll see that I turned sharpness down completely to compare, and then over and over people said that was the wrong thing to do. Even though I think (as you do) that it was exactly the right thing to do.

So, to satisfy those who complained about it, you'll see in the first post here that I used the stock/default "landscape" setting, which includes sharpness at +3.

And so along comes Chris_TC to say "what's with the sharpening?"

Sigh... :)

Chris_TC
10-16-2009, 09:22 AM
Chris, look at my other thread, you'll see that I turned sharpness down completely to compare, and then over and over people said that was the wrong thing to do.
Yes, I read that other thread after I had posted here. And I stated (http://www.dvxuser.com/V6/showpost.php?p=1779846&postcount=454) in that very thread that you had put together a great test, thanking you for not using in-camera sharpening.

Anybody who claims you should sharpen in camera simply doesn't know what they're talking about. Those double contours jumped out at me immediately, reminding me of the bad old days of edge "enhanced" DVDs.

MovieSwede
10-23-2009, 09:54 PM
Anybody who claims you should sharpen in camera simply doesn't know what they're talking about. Those double contours jumped out at me immediately, reminding me of the bad old days of edge "enhanced" DVDs.

Bad old days? Have you seen the Gladiator BD. ;)

Mitch_Ives
10-27-2009, 12:21 PM
Canon said that they know what we want. They said that it's hardware that's the difference. And they said that they'd build it, and that it'd be more expensive.

Anybody else, besides me, that would be okay with that?

Lammy
10-27-2009, 12:37 PM
oki, so the solution wud b 2 mek a custm firmware to hack the photo settingz and processor bioz to only take 1 megapixel over da whole sensor, rather den 14 megapixels, den mek it so liveview is tweaked 2 operate lyk, natively wit a better debayering algoritm, and all sharpning, detail loss n moire roblems sorted!!!1111one