Panasonic AF101 Footage

tflak said:
Compared to everyone else here I don't know squat, so forgive me if what I say is just plain naive. But you speak as though this were a finished product. Or is it the case that the things you're describing couldn't possibly be different in the shipping model? Just curious.
Click to expand...

The two main features are set in stone and these define the camera target market. It's not possible than these characteristics could change:

1) The rolling shutter is visible in the samples and is an artifact of the current CMOS generation. Almost everything in the CMOS world has it. If it was global shutter we would know from Panasonic marketing and a rock solid frame would be visible in the samples. Rolling shutter is not something that changes from one firmware version to the next. A new sensor design is required, a more expensive design, a complex one, and one that adds its own compromises. AF-100 is not a global shutter CCD HPX500 or HPX3700.

2) The GOP compression codec is a given. We have seen variants of these codecs. AF-100 does not have an intraframe codec like the higher end Panasonics. It's not DVCPRO HD, it's not AVC-Intra. Adding P2 doubles the cost of the camera.

Also, I don't see why this sensor would be better in low light, have higher sensitivity and lower noise compared to sensors of the same generation and pixel size in other Panasonic cameras. I haven't seen Panasonic marketing mentioning anything in that department. Did someone from Panasonic say it will be 2 stops better in noise or something like that? The noise visible in the samples is normal for this sensor, the still camera has a similar sensor and similar noise. Signal/noise is affected by technology and pixel size. One of the two needs to change drastically in order to provide better performance. One can change firmware to process more, and reduce noise using numerous techniques, but that's not true image quality, that's denoising in camera and always comes with artifacts. A softer image or worse.
 
Otis Grapsas said:
The highlight behaviour on the face at 1:00 (false magenta tone and then clipping to white) is because of the ITU709 gamma curve. The camera probably has better settings. If it has a CineV curve like the HVX, it will clip a lot better, with less false color. I wouldn't call it filmic, but a lot better than ITU709 in highlight behaviour and also midrange contrast and color.

Alias appers to be reduced or eliminated. I believe the fabric at 1:03 would alias on a typical dSLR. It would be visible on 720p Vimeo.

The rolling shutter looks identical to the still camera. I can tell even on very slow camera motion and vertical panning at any speed. Red One fools me sometimes, but only in some very slow moves.

At one point, I read that the noise in the shadows is because of video compression and was not on the original clip. That's ridiculous. Compression cannot add noise of this type and at this level, it can only change the character of input noise, or exaggerate it to a degree. The camera was on high ISO probably. At 2:45 one can see that Vimeo preserved the camera noise in the hair shadows. It's the typical bayer CMOS through GOP compression look of noise. With so little motion, even Vimeo compression can encode noise if it's high enough.

All in all, I believe the camera preserves the still model image quality in rolling shutter, sensitivity and noise but adds a better codec, antialiasing filter, professional menus, audio, professional inputs/outputs etc. One cannot add all that to a dSLR, but adding half of it will take the cost pretty close to the AF-100 cost. It's definitely a good buy but don't have unrealistic expectations. It solved all other problems, but it will still perform to rolling shutter CMOS, medium pixel size noise/sensitivity, and GOP compression level, because it's an affordable camera.
Click to expand...


Do you own a HDSRL? I see an extreme reduction in jello during the closeup shots of her against the window when there is camera movement compared to the HDSLR.
 
I did not notice jellow at all.....My concern is the noise in the blacks ....
I'm hoping its cleaner then the HMC 150.

Just gonna wait for some raw footage of the final production model .
 
This sensor (used in GH2 and AF100) is better in low light.
It had been stated in GH2 release and on site materials.
I can tell with pictures I have and footage I have.
According to Panasonic noise measurements (for stills modes) is is from 1.5db to about 3.5db better at different ISO values.
Movie can have even slightly bigger advantage. It all depends on implementation.


As for jello.
Otis has some fix on this.
I don't even know that to say.
 
im making an educated guess here but heres my assumption basing on reports and theories
so my best guess is its very good at 1600 ISO.

5d mkII is at its limit at 1600iso in acceptable noise. so thats your bar for comparison.

jello wise, if its not d90 jello its good. anything equal or higher than 5d mk II jello, is acceptable.
 
Hi Guys,

I just don't see the jello in the camera I have nor in the footage shared, look where there is movement against the windows, there is no skew. And just remember the camera is not finished yet.

Best,

Jan
 
Jan_Crittenden said:
I just don't see the jello in the camera I have nor in the footage shared, look where there is movement against the windows, there is no skew. And just remember the camera is not finished yet.
Click to expand...

I fully agree that jello is reduced compared to GH1.
CMOS readout speed is much faster. And here we see results.
But I don't think that we'll see any changes in readout speed (and jello as consequence of this) in final version.
Sensor changes of this scale are almost impossible at this stage.
 
Last edited:
can someone tell me why GH2 footage I've seen so far looks absolutely terrible but AF100 footage looks awesome and they have same SENSOR
 
Homunculus said:
can someone tell me why GH2 footage I've seen so far looks absolutely terrible but AF100 footage looks awesome and they have same SENSOR
Click to expand...

First. GH2 had been provided to many guys, most of them are unable to shoot something really good by nature. They are testers, after all :-)
Second. AF100 footage had been carefully shot. With good lenses. And I am with Jan on this. It is much better to be this way.
Third. Codec can be very different. For example, in motion prediction part. I am looking at high detailed water flow footage made with GH2 with 0.2 firmware.
And I do not like that I see. Looks like extremely compressed HD satellite feed. Better than GH1 but not really much.
 
Vitaliy Kiselev said:
First. GH2 had been provided to many guys, most of them are unable to shoot something really good by nature. They are testers, after all :-)
Second. AF100 footage had been carefully shot. With good lenses. And I am with Jan on this. It is much better to be this way.
Third. Codec can be very different. For example, in motion prediction part. I am looking at high detailed water flow footage made with GH2 with 0.2 firmware.
And I do not like that I see. Looks like extremely compressed HD satellite feed. Better than GH1 but not really much.
Click to expand...

man I feel bad for gh2. it's going to be the ugly red headed step child of the panasonic line up. gh13 is still better and af100 is much better than both so gh2 is going to be forgotten about. maybe they should just cancel gh2 and save themselves the $$$.
:furious3::embarasse
 
PaPa said:
Do you own a HDSRL? I see an extreme reduction in jello during the closeup shots of her against the window when there is camera movement compared to the HDSLR.
Click to expand...

I don't own a dSLR because I never bought the SLR for video. I have experience with almost all of them, most common CMOS camcorders, and a series of CMOS sensors with rolling and global shutter.

?You can compare the performance of two different rolling shutters by memory of different situations using lenses of different focal length in cameras of different weight and size with or without stabilization, handheld or an a tripod, at different shutter speeds, and you are sure that you can see an extreme reduction? Congratulation! I'm not that precise! The rolling shutter looks normal to me, it could be 10-15% slower or faster, that's not that important, all I know is the frame is unstable on vertical motion and this is no global shutter. I have enough experience to not have to wait for skew or jello to appear to know there is a rolling shutter. Many professionals can do the same and can spot the rolling shutter even when there is minimal camera motion or simply some camera instability. They could be insane but they could also be more perceptive to the phenomenon. I can also spot CCD smear without visible smear on the screen because the smear is always there. If you examine parts arounds an overexposure there is a glow that does not exist in CMOS sensors. I never got attacked for spotting smear though:)

If the rolling shutter is faster, Panasonic can quantify how much faster it is. Or we can wait to see how fast it will be. When a sample from a new CMOS camera appears many people want to believe the rolling shutter problem is not there even when it exists to a degree on the very sample. When the camera is released, the rolling shutter is always there. What is more logical, assuming the obvious based on facts of the past models or assuming the problem is "extremely reduced" after seeing a couple of shots shot by somebody else and released after permission by the company and under very tight control? Early samples never show the flaws of a camera, isn't that a given in marketing?
 
Vitaliy Kiselev said:
As for jello.
Otis has some fix on this.
I don't even know that to say.
Click to expand...

You should address me in person when commmenting my posts. You know I'm following the discussion, right?

You agree "camcorder" AF-100 and "still camera" GH2 will probably be equal in low light, so why do you present it as a disagreement?
 
Is it just me and my lack of technical knowledge or is this discussion headed in a "If a tree falls in the forest..." direction? I normally leave the room at that point.
 
Otis Grapsas said:
I don't own a dSLR because I never bought the SLR for video. I have experience with almost all of them, most common CMOS camcorders, and a series of CMOS sensors with rolling and global shutter.
Click to expand...

Ouch.

If the rolling shutter is faster, Panasonic can quantify how much faster it is. Or we can wait to see how fast it will be. When a sample from a new CMOS camera appears many people want to believe the rolling shutter problem is not there even when it exists to a degree on the very sample. When the camera is released, the rolling shutter is always there. What is more logical, assuming the obvious based on facts of the past models or assuming the problem is "extremely reduced" after seeing a couple of shots shot by somebody else and released after permission by the company and under very tight control? Early samples never show the flaws of a camera, isn't that a given in marketing?
Click to expand...

Otis, I consider you as special "feature" of this particular forum :-)
So, at least in virtual discussions about non existing products you must be accurate :-)
I even can tell you that GH2 rolling shutter is 2 times better than GH1 rolling shutter (and AF-100 uses same sensor, so you got me, right?).
Is this enough to make new entry in your book for "failed" CMOS sensors?

I find this stupid talks about rolling shutter so entertaining :-) Just most readers do not think so. :-)
 
Otis Grapsas said:
I don't own a dSLR because I never bought the SLR for video. I have experience with almost all of them, most common CMOS camcorders, and a series of CMOS sensors with rolling and global shutter.

?You can compare the performance of two different rolling shutters by memory of different situations using lenses of different focal length in cameras of different weight and size with or without stabilization, handheld or an a tripod, at different shutter speeds, and you are sure that you can see an extreme reduction? Congratulation! I'm not that precise! The rolling shutter looks normal to me, it could be 10-15% slower or faster, that's not that important, all I know is the frame is unstable on vertical motion and this is no global shutter. I have enough experience to not have to wait for skew or jello to appear to know there is a rolling shutter. Many professionals can do the same and can spot the rolling shutter even when there is minimal camera motion or simply some camera instability. They could be insane but they could also be more perceptive to the phenomenon. I can also spot CCD smear without visible smear on the screen because the smear is always there. If you examine parts arounds an overexposure there is a glow that does not exist in CMOS sensors. I never got attacked for spotting smear though:)

If the rolling shutter is faster, Panasonic can quantify how much faster it is. Or we can wait to see how fast it will be. When a sample from a new CMOS camera appears many people want to believe the rolling shutter problem is not there even when it exists to a degree on the very sample. When the camera is released, the rolling shutter is always there. What is more logical, assuming the obvious based on facts of the past models or assuming the problem is "extremely reduced" after seeing a couple of shots shot by somebody else and released after permission by the company and under very tight control? Early samples never show the flaws of a camera, isn't that a given in marketing?
Click to expand...

Being 10% or 15% faster or slower IS important. You are making claims about the sensor and rolling shutter, yet clearly, you've not really done your homework regarding the AF100. It is NOT the same sensor as the GH1, so clearly it is possible for it to have different characteristics. Also, pertaining to read/reset, you are absolutely wrong. In some CMOS sensor designs, you can actually adjust the read/reset in firmware. For instance, the RED ONE MX sensor had it's read/reset set faster then slower with subsequent firmware releases. Granted, this assumes the clocking structure of the CMOS circuitry is not hard wired to a clocking frequency. Depends on the design. But not impossible.

More than likely, most CMOS sensors are dialed back based on the speed variance of the CMOS sensor process variation so that they get a higher yield of usable sensors to keep the cost down. I am a CMOS engineer, so I have the background to understand CMOS manufacturing and chip design as a whole (this is my day job).

Have you tested the AF100 prototype?

Barry Green has tested both aliasing/moire, rolling shutter, and dynamic range of the sensor as much as he could on the prototype. All three things he said were better. Granted, I am sure he'd have liked to give examples, but he wasn't allowed. Obviously I will be doing tests compared to my 7D when I get it, and not doing them subjectively either. It will be very clear as I measure the skew characteristics as I want to know how much better it is.

You do know that film also has skew right? Granted, film doesn't have the hard edge issue of CMOS reading since the spinning shutter will impart a diffused effect over the edge as it passes over the film, it still happens. I've seen film skew on Saving Private Ryan, Bourne Identity, and a few really old films. And when I say I saw it, I actually brought the frame grab into photoshop so I could analyze it. It is there.

And CLEARLY not all rolling shutter is the same. My iPhone has the worst rolling shutter of any camera I own! :) And the Nikon D90 is a close second. Getting that read/reset as close to film skew as possible will at least give people something they are used to. Granted, I'd love a global shutter as well, but I would not want the active area hit on the pixels.

Also, even within an already designed sensor, processing technology can get better to make a better sensor as well. CMOS has quite a bit of process variation and temperature dependence. I am amazing at what imagers do these days given that!
 
I don't see any jello either. I general I love the footage, it looks different from what other pro videos cameras produce, but I wouldn't say it looks worse, it just looks different and I prefer this look to the look of more expensive video camera's. I would call it the DSLR video look with all the disadvantages of DSLR video taken away, and I simply love it. I trust Panasonic on doing the best they can to make it a proper final product but in the end there will always be people who don't like the look and would be better of with something else.
 
Homunculus said:
man I feel bad for gh2. it's going to be the ugly red headed step child of the panasonic line up
Click to expand...

lol. The GH2 will come in black and silver only, not in red :-)

I pre-ordered one to use as B camera with my AF101. Of course it will be better than the GH1.
 
Honestly, all the 7D was missing is all of the things that the AF100 is bringing to the table. I was quite pleased with the image quality, but wanted to sync good quality sound to the camera. I like implementing camera movement whenever a scene calls for it, but if the lens was too long and I wanted a hasty handheld, I would have to resort to using wider lenses to avoid seeing intense skew. Deep focus shots were riddled with Aliasing and moire, which made it impossible to have any type of camera movement without throwing the image slightly out of focus and then sharpening in post.

Now if the AF100 delivers on those respects and can still provide an image as nice as the HDSLR's have been producing, disregarding that we are gaining another stop to two stops of latitude (suggested), then I think the AF100 is a total hit and did exactly what it was sent out to. The image quality/ compression isn't worse than what the 5D, 7D or GH1 have been producing is it? Cause if it's even a bit better then another sweet plus.

The pre order is done with BandH and I guess the real test will be when those of us who ordered one receive them and learn all about them inside and out.
 
Vitaliy Kiselev said:
Ouch.



Otis, I consider you as special "feature" of this particular forum :-)
So, at least in virtual discussions about non existing products you must be accurate :-)
I even can tell you that GH2 rolling shutter is 2 times better than GH1 rolling shutter (and AF-100 uses same sensor, so you got me, right?).
Is this enough to make new entry in your book for "failed" CMOS sensors?

I find this stupid talks about rolling shutter so entertaining :-) Just most readers do not think so. :-)
Click to expand...

Vitaliy, your smileys are trying too hard:)

I never mentioned GH1 but you keep repeating GH1. The point of Af-100 is to provide more power vs the consumer still model which uses the same or similar sensor, in this case, that would be the GH2. That's its power, inputs, outputs, control on the image and the camera, a better recording format. I would be surprised if Af-100 is better than GH2 in rolling shutter, low light and sensitivity. I'm puting the camera into perspective and you see it as an attack to the product. This leads nowhere.
 
You must log in or register to reply here.
Back
Top