af100 vs 5D mk ii

relatively speaking the aliasing and skewing is extremely minimal on AF100. HOWEVER i have said for months that i will gladly take a camera with moire/aliasing that has deeper dynamic range as you have just said. The reason is that it is my personal opinion that dynamic range is the most important final element missing from these digital cams in terms of achieving a filmic look. i.e. it is the biggest gap left. Moire can be easily avoided, aliasing to some extent can as well though it is never a major problem to begin with. Skew can quite easily be avoided and is almost a nonexistent problem. But how can you avoid poor dynamic range? It will be evident everywhere that your latitude sucks!
Keep in mind this rant is not to say that 5d has higher dynamic range than AF100, I highly doubt that it does.
I'm sure AF100 has at the LEAST as high dynamic range if not slightly higher, but I am extremely interested in finding out HOW much higher? Supposedly 5dmkii has 8.5 stops and AF100 10.5 stops but from the comparisons I've seen so far it doesn't look like it's 2 stops higher, really interested in someone testing this
 
Sorry, I was away from my computer for a bit.

From my tests on the XHA1/T2i/AF100 as well as our shoot today which incorporated both the AF100 & XHA1 I would say:

- The detail and resolution on the AF100 are top notch (and REAL DETAIL to boot)
- Color and "mojo" are all pretty easy to match between all three cameras
- Noise is NOT A PROBLEM (and I was one of the people quietly panicking)
- There is VERY little as far as moire and aliasing coming out of the AF100, I am seeing much more out of my XHA1 and about 1,000,000,000 times more on my T2i (Scientific..I know)
- Skew I cannot comment on as I haven't run into the need for quick pans yet.

*Caveat - My tests and results are NOT scientific in nature...I am using my eyes and at most 100%-200% viewing...you want science, talk to one of our many forum gurus.
 
Homunculus said:
I'm sure AF100 has at the LEAST as high dynamic range if not slightly higher, but I am extremely interested in finding out HOW much higher? Supposedly 5dmkii has 8.5 stops and AF100 10.5 stops but from the comparisons I've seen so far it doesn't look like it's 2 stops higher, really interested in someone testing this
Click to expand...

Me too. I'd want it done by a third party though... here the AF-100 would be the winner and on a Canon site... the 5D. Dynamic range can me measured too subjectively. I'd love to see a Zacuto "Generation II" shootout though. Even though Canon's "generation II" is looking pretty poor to non-existent at the moment.
 
Hywel said:
Unfortunately, that hard clipping is an inherent limitation of digital sensors. When the electron content of an individual pixel gets to a certain maximum level (called "full well"), shining more light on the sensor does not produce any extra signal. This is unlike film, where exposing over-exposed film did still give a little extra signal, up to several more stops, which is what produces the film-like soft blowout we know and love.

To combat this in-camera, all one can do is to underexpose the image, protecting the wells from filling up, and then turn the gain up on the mid tones to compensate for the under-exposure. This amplifies the noise, and it is a tradeoff made by the camera designers how best to trade off between the highlight roll-off and the gamma curve to produce acceptable pictures.

RED's HDRx solution to this is very clever, incidentally- to produce a second, shorter exposure in the time when a film camera would be moving the film, and record that second data stream alongside the original, allowing one to recover highlight detail in post because you have the "right" exposure and a suitably underexposed frame to blend together in post.

The only other way to deal with this is to make sensors with larger full-well capacity (thus increasing the light level at which blowout occurs) and lower noise (thus increasing the amount the signal can be boosted in the mid-tones before the noise becomes unacceptable). As heat is a primary source of noise, this is why so much of the Alexa is devoted to cooling (and why Astronomical camera chips are almost invariably cooled, because they are looking at very small signals over long integration times to beat down the noise).

At the AF100's price point, the sensor was never going to be as low noise as the chip in an Alexa. The different gamma curves give you some control over where the noise/rolloff tradeoff is to be made, at least. If you are in very controlled lighting conditions, you can use the cine-like curves which do relatively little to the signal from the chip, but if you are in an uncontrolled contrasty or changing light situation you may want to protect your highlights and live with some moderate increase in noise. If using the cine-like curves, you should underexpose by half a stop, according to the manual, which puts the mid-tones in about the right place on the curves, by the way.

Cheers, Hywel
Click to expand...

Thank you for your basic overview of digital image sensors... maybe you dont understand what I'm saying?

We are talking about the AF100. Yes there are other cameras with better sensors. Yes digital image sensor photosites all have a given photon well capacity and a photosite blows out absolutely when it is full. I was discussing how to avoid the making these blow outs look ugly with a gamma curve.

HDR has been around for quite a while, and HDRx doesnt really help any of us who do not own a RED EPIC-M which are the only cameras currently "available" that feature this technology.

I dont think anyone was expecting Panasonic's $4k+ camera to match the latitude or noise or resolving power of Arri's $50k+ new camera that uses a larger sensor and a unique purpose-built dual gain architecture. I was, however, expecting or at least hoping cine-like curves to be a little cine-like (at least attempt to roll off highlights, at least attempt to be optimized for highlight latitude).

What do you mean when you say cine-like curves do relatively little to the signal from the chip? Is there something I'm missing? Isnt it just another gamma curve that essentially completely remaps the entire linear response of the chip into a gamma-corrected, viewable image?
 
Homunculus said:
just to give an example, watch the footage of af100 compared to 5d of the cars on the race track (that older footage). AF100 looks ten times better and more filmic with the landscapes, cars, etc whereas 5dmkii looks video-y and seems to have poor dynamic range.
but then footage showing actual people and skin tones, the comparison gets much closer
Click to expand...

Really? You mean the one where the AF100 consistently had harsh, distracting blowouts on the hood of the car?
 
nyvz said:
Really? You mean the one where the AF100 consistently had harsh, distracting blowouts on the hood of the car?
Click to expand...

believe me I have been infamously one of the biggest critics here of the af100's nasty clipping but that video did not seem to exhibit that to me. its rendition of the white car was far better than the canon footage of the car which was completely blown out
 
Homunculus said:
relatively speaking the aliasing and skewing is extremely minimal on AF100. HOWEVER i have said for months that i will gladly take a camera with moire/aliasing that has deeper dynamic range as you have just said. The reason is that it is my personal opinion that dynamic range is the most important final element missing from these digital cams in terms of achieving a filmic look. i.e. it is the biggest gap left. Moire can be easily avoided, aliasing to some extent can as well though it is never a major problem to begin with. Skew can quite easily be avoided and is almost a nonexistent problem. But how can you avoid poor dynamic range? It will be evident everywhere that your latitude sucks!
Keep in mind this rant is not to say that 5d has higher dynamic range than AF100, I highly doubt that it does.
I'm sure AF100 has at the LEAST as high dynamic range if not slightly higher, but I am extremely interested in finding out HOW much higher? Supposedly 5dmkii has 8.5 stops and AF100 10.5 stops but from the comparisons I've seen so far it doesn't look like it's 2 stops higher, really interested in someone testing this
Click to expand...


I don' t know where this number of 8.5 stops dynamic range of the 5d come from. There is only one site on the internet where I have seen consistent test for the video cine camera and it is provideocoalition, you can see it test of the 5d dynamic range here http://provideocoalition.com/index....d_how_much_dynamic_range_does_it_have_really/ which shows more in the 10+ stops. It has test of the red mx and alexa also if you want to have a good idea between camera.

Until now the Af-100 dynamic range does not look good at all. We have been told since the Japan trip fiasco that it was much better but I am still to see that. Things seem to blow even in overcast dull weather. Of the two, I have been more impressed by the gh2 in that respect. It is as if the video division is still working with the old video mentality of low dynamic range. People in the video world have been use to work for so long with low dynamic range compared to photography.
 
Homunculus said:
believe me I have been infamously one of the biggest critics here of the af100's nasty clipping but that video did not seem to exhibit that to me. its rendition of the white car was far better than the canon footage of the car which was completely blown out
Click to expand...

I don' t remember the one with the canon, but the one of the af-100 was blowing out on the hood of the car even if the rest of the image was under exposed.
 
Thomas Church said:
I have to say that I've watched every single AF100 video posted online, and so far there hasn't been a single one that has the "mojo" that 5D footage has.

Yes, it has less technical problems and so on, but the footage doesn't "pop" or look as filmic as 5D footage does.

I'm hoping that someone can create a Scene File that truly makes the AF100 sing.

Because at the moment I'm not very tempted to get one.
Click to expand...


gustavo_viral said:
Im an AF100 enthusiast and i bought one. But i have to say that i kind agree whith this...
Click to expand...


Thanks for saying that. I thaught I was going crazy and on other threads I nearly dot insulted for daring to say that.
I'm not a AF100 owner but I agree, I haven't seen great footage yet. And simple images seems to have more inpacts on me when its a 5D.
I think its a great camera but for short films and just image wise I doubt it has that "mojo" thing as you say. And I really wished it did, It would really really make my life easier. But I don't whant to go back to that video feeling since today, in my every day work with loo budget, I'm reaching that cinema like feeling
 
To the people who play down the moire/aliasing: Whenever I had it in my shot the client would point at it. Mostly still accepting the product, but still they SEE it.
On the other hand I have yet to get a client complaining about the noise in my HVX....(although I admit we find it disturbing in extreme low lights). This is exactly what Panaonic does: the overall image is pleasing and has no real visible flaws (for its price-point). And don´t forget the camera is workable: audio, waveform, peaking, HD-SDI output.....there´s not much left for the 5D to do in my world.....

EDIT: again look at the pricepoint. The 5D is not cheap for what it does. The AF100 is no Red or Alexa. It is a grown-up DSLR at a crazy pricepoint with a much more flexible mount than Canon (I finally can use my MD mount lenses again).
 
The thing is with video, as frustrating as the lack of a dynamic range is at times, expose within the limits imposed on you using waveform, zebras and external monitor,
and you have a lovely image on the HVX, AF100 (not got mine yet, but see no reason why it's not more of that same mojo loveliness from Panasonic from what I've seen shot).
Rather like shooting transparency film in the good 'ole days.

The thing is with aliasing, moire etc is that if you have it on a shot then you're pretty much f*cked, as there is not much filtering and the like that can be done to rescue it without even further degrading the image.
I'm talking on a top-notch Nuke/Shake level here, with 16-point sampling, bicubics etc.

I worked on the post on a top commercial last year with lots of cityscape stuff shot on a 5D cut in with HD, 35mm etc,
and with subtle pans across Cityscapes there was so much aliasing and moire from the 5D stuff that a lot of the comping with the 5D stuff was tracking in still image patches over the offending areas.
All a total 'mare timewise, as there was not allocated time for these fixes. Hated it, especially also having to camera track the lovely jello-footage to get all these patches in.
And then after all that malarkee we were then able to get onto the comping...

I was watching the BBC news last night. Usual presenter stuff, nothing special, and then it came to the weather, and the girl there was dressed in dark blue (against bluescreen),
The seperation between her and the background was so bad that she was buzzing all over the place because of the hard key, and the aliasing that introduced.
Very distracting.

And something I find totally unaceptable for a so-called hi-end image.

just my 2 cents.
 
I had a play with the new Panasonic AF101 last night. I still think it looks pretty odd but it's a very nice camera.

It takes a lot of features from the HVX200 so if you know your way around that camera then you'll quickly get on with the AF101. It is very small and light but still has the ergonomics of a video camera. Meaning there are all the controls you could need on the side plus the inputs and outputs of a pro camera.

I also had a Canon 5D MkII and a Lumix GH2 (Panasonic's new DSLR with HD video capabilities). My main purpose for having these cameras was to compare them and discover their aliasing properties. One of the main complaints of the 5D is that it has pretty awful aliasing issues so I was asked to see how it compared to these other cameras.

My totally unscientific test involved filming something with lots of horizontal lines - a Lacie hard drive. I tried to set the cameras up in a similar way and used the same focal lengths.

The results were a little damning for the Canon 5D MkII.

Here's a link to a still (it's a compressed but large jpeg) and it's pretty conclusive as to which is the worse camera for aliasing.

https://files.me.com/smokescreen_tk421/1y5zxq

The GH2 and the AF101 share a lot of components so it is no surprise their images are similar.

(Disclaimer: I would like to add that this was a totally unscientific test and I've not operated the 5D too often so I might have missed a setting or something. And also I am very biased towards Panasonic cameras and I hate DSLRs.)
 
nyvz said:
What do you mean when you say cine-like curves do relatively little to the signal from the chip? Is there something I'm missing? Isnt it just another gamma curve that essentially completely remaps the entire linear response of the chip into a gamma-corrected, viewable image?
Click to expand...

Yes, but the point is that they do so less aggressively than the other gamma choices offered by the camera (at least, assuming they are doing the same as the equivalent settings on the HVX200). That gives a flatter, more linear response but at the expense of no headroom at all when it comes to highlight clipping. You can EITHER have highlight protection, or lower noise and better mid tones- but you have to pick. Asking for a soft rolloff in the highlights of a digital sensor (which is what I understood your original post to be asking for) is not a no-consequence request. The only way to get it is to underexpose (to avoid clipping) and gain up the mid-tones.

The Cine-like curves in Panasonic camera have traditionally left the highlight protection in the hands of the user, in order to get the best result in the rest of the tones. Net result: it looks great, nice dynamic range, low noise, probably the cleanest and best image you can get out of the camera. Right up until the highlights start to clip, at which point they clip in a spectacularly ugly way... because the way they've got the best out of the sensor is NOT to implement the shoulder used by the other gamma curve options in the camera.

So these curves have a more linear response, without significant shoulder. They must be used with care (as I said, they are really for controlled lighting conditions). When used in this way, they generally get the best from the camera, but at the expense of the removal of highlight protection.

Ironically, the look you get out of the camera with Cine-like gamma is not very cine-like. It is flat and suitable for grading, with good dynamic range, which lets you apply a gamma curve in post to get an image with as soft a roll-off as you like, just so long as you haven't clipped. It does NOT have a film-like rolloff in the highlights- it just has the information in the footage to let you impose that in the grade. So long as you haven't clipped!

One gets quite religious about the 100% Zebra shooting in Cine-like mode :-) I'm very much looking forward to the waveform monitor on the AF100, which will help balance the lighting so the blacks are where you want them to be as well (dark, but retaining some shadow detail, so you can choose how far to crush in post, too).

Stu Maschwitz's DV Rebel's guide book has an extensive section on cameras on its DVD which goes into this in some detail for the HVX200 and it sounds like this carries over to the AF100.
 
Last edited:
I had the Canon 5d MK II until my last project for a big car parts manufacturer in Germany. There were several (very important shots) that were absolutely ruined by the aliasng the 5D produced!

In general I am of course still pleased with the 5D-Magic but the lack of 60p (at least 720p) , jellow, aliasing, moiree etc ......made me take a look at the GH2! hoping, it will feature 1080p60 or 50p, having less issues with jellow and aliasing!

The shock came, when I heared it doesn´t have at least 25P in the lovely 24Mbit AVCHD.

In 24p the jellow is as bad as the 5d, although I have to admit, that the GH2 pictcture-quality sometimes seems to be like an updated or cleaned 5D MK II(from aliasing).

Last but not least, there seems to be not better solution (for the price) than the AG-AF100 or AF101 for me - always 24MBIT , 1080p60 , ETC...
 
Gary Askham said:
I had a play with the new Panasonic AF101 last night. I still think it looks pretty odd but it's a very nice camera.

It takes a lot of features from the HVX200 so if you know your way around that camera then you'll quickly get on with the AF101. It is very small and light but still has the ergonomics of a video camera. Meaning there are all the controls you could need on the side plus the inputs and outputs of a pro camera.

I also had a Canon 5D MkII and a Lumix GH2 (Panasonic's new DSLR with HD video capabilities). My main purpose for having these cameras was to compare them and discover their aliasing properties. One of the main complaints of the 5D is that it has pretty awful aliasing issues so I was asked to see how it compared to these other cameras.

My totally unscientific test involved filming something with lots of horizontal lines - a Lacie hard drive. I tried to set the cameras up in a similar way and used the same focal lengths.

The results were a little damning for the Canon 5D MkII.

Here's a link to a still (it's a compressed but large jpeg) and it's pretty conclusive as to which is the worse camera for aliasing.

https://files.me.com/smokescreen_tk421/1y5zxq

The GH2 and the AF101 share a lot of components so it is no surprise their images are similar.

(Disclaimer: I would like to add that this was a totally unscientific test and I've not operated the 5D too often so I might have missed a setting or something. And also I am very biased towards Panasonic cameras and I hate DSLRs.)
Click to expand...

Thanks for testing.
Is it me or the gh2 jpeg test looks cleaner than the af100?
Also, why all af100 footage I see has this yellowish image color?
 
roei z said:
Thanks for testing.
Is it me or the gh2 jpeg test looks cleaner than the af100?
Also, why all af100 footage I see has this yellowish image color?
Click to expand...
i have seen the yellow too but in this particular one the canon is the really yellow one
 
I didn't spend too much time setting the camera's up - I maybe should have white balanced properly and spent more time going through the settings until I was sure the shots would be as similar as possible. But still.

I was impressed with the picture quality of the GH2 but the ergonomics are wrong for a video camera. In fact it's so small it's almost no good as a stills camera either. I have pretty big hands and can only fit 3 fingers on the grip.
 
You must log in or register to reply here.
Back
Top