Getting to the bottom of the noise complaints - is it h.264 uncompression software?

[Barry_Green]

Or, the simple solution is just to take footage that was shot both internally and externally and look at the noise.

Well, that's the problem -- I don't see the noise! If I had a situation that showed noisy footage I could then do exactly what you suggest. But the footage looks awfully good over here, so I don't know what people would say is noisy!

Does the chart look noisy to anyone? How about the color bars?

I'm halfway to the point of thinking that Leo's footage is just suffering from some curse, and throwing it out of the mix and replacing it with something else.

So, yes, let's start over with this: anyone who has seen noise -- look at the chart. Is there anything in the chart footage that you would describe as "very noisy" or "unacceptably noisy"? If so, I can re-shoot that with the Nano.

 

[HHL]

Well, that's the problem -- I don't see the noise! If I had a situation that showed noisy footage I could then do exactly what you suggest. But the footage looks awfully good over here, so I don't know what people would say is noisy!

Does the chart look noisy to anyone? How about the color bars?

I'm halfway to the point of thinking that Leo's footage is just suffering from some curse, and throwing it out of the mix and replacing it with something else.

So, yes, let's start over with this: anyone who has seen noise -- look at the chart. Is there anything in the chart footage that you would describe as "very noisy" or "unacceptably noisy"? If so, I can re-shoot that with the Nano.

Barry....when you look at this image above...(click it twice so you see the full version)....do you see the "missing blocks" on the inside of her right arm?

 

[HHL]

Here it is at 200%. (I took the image and resized it by 200 percent)



Here's another frame: (200%) Notice how the missing pixels (noise) has all moved...but still just as bad. This literally makes it "dance"...so her arm is distorted. It's impossible to miss on my screen. Do you see it?

 

[HHL]

Hey Guys! Will someone who's image looks good please capture (.jpg 200% resize) this same area to validate that pixels are ACTUALLY NOT MISSING on their version. This obviously isn't calibration material....this is a question of "short cutting" the decompression as Barry is implying. Anyone up for this?

 

[Barry_Green]

I see some noise in that shot, yes. But nothing that makes me grab my hair and scream "oh no, it's horrible!" It just looks noisy -- which I would expect for a 7-candle shot.

But, seeing as this footage is causing no end of grief, and seeing as I have no scientific "control" over it, I'm calling a do-over -- I'm booting the candlelit shot because I don't want it to become the focus of everything everyone's looking at.

I'm starting over with a "still life". This one I have direct control over as I just shot it.

So no more candlelit shot, no more Leo's footage, no more uncontrolled variables. Let's go with a much more scientific approach, working with footage that is a known quantity. For those curious, yes this footage is shot on THE EXACT SAME PROTOTYPE CAMERA that Leo's footage was shot on. But at least now I can verify everything about it, from the lenses to the settings to the white balance, so it is an entirely known quantity with no mysteries about it (and the Leo footage is mysterious -- the candlelit and bellydance clip didn't play on my blu-ray player! I don't know why.)

New footage card is uploading right now, and I would like to use it as the basis for all comparisons.
http://www.mediafire.com/?wzdcba84rbaj9tg

 

[Taylor Rudd]

Here is the "March of the Ants" as Barry called it. Left side...many many blocks that have an almost yellow hue to them.
View attachment 27555

Can you see this...or has it been lost in the posting of it here?

BTW...this is the same size as it appears on my screen. I haven't zoomed in or anything...I'm not trying to accentuate it at all. This is just "as is". Believe it or not...it actually looks "better" after taking it into Photoshop and exporting it at 100% quality...it crushed it a bit (in .jpg)..which helped some. (Believe it or not)

I can see blocks of noise if I turn my backlight up all the way and tilt my screen down. No doubt about it, there is junk in the blacks...but it isn't something I would see if I were looking at an appropriate screen brightness and viewing angle.

I trust the pros when it comes to talk of encoders/decoders. I can force myself to see funk in the blacks, but it isn't visible in a natural viewing situation.

Macbook Pro
15" LED
Calibrated with Spyder2Express, backlight at 87.5% (14 out of 16 clicks)

 

[timbook2]

I got the new file finally. OSX 10.5.8 powerbook 2.66 QT 7.
VLC looks terrible but we know that
I used Voltaic to convert the .mts files to and see no armies of ants BUT: in clip 002 I see Barrys arm turning the cam on+off and lots of strange reflections which dont help distinguishing noise in the greys
In clip 000 are you inserting the ND filter? and closing/opening the iris ? the shadow of the monsters hand across the girl in pink ( who now does this indian sitcom) looked like a hair in the lens...
no noise for me so far, will check FCP transcode later

 

[HHL]

ZERO NOISE....PERIOD. As I have observed in all of the other AF100 footy. What ISO here Barry? (For documentation purposes only here)

 

[TimurCivan]

Damn if I had a AF, I could put together something with exact exposures. -5stops to +5sttops, and we can compare the grain structures at each exposure level.

Also, what ISO are you @ barry?

 

[Kit Hannah]

I still think it has a lot to do with people's monitors. The stock settings for the Nvidia cards have the gamma way too high, bring it down and much of the "noise" goes away. My laptop has settings for overall computer and video only.

 

[Barry_Green]

I used Voltaic to convert the .mts files to and see no armies of ants

Interesting -- do you see armies of ants when not using Voltaic? You see them with VLC, obviously... so VLC is definitely contributing to it, but Voltaic isn't?

BUT: in clip 002 I see Barrys arm turning the cam on+off and lots of strange reflections which dont help distinguishing noise in the greys

Yes, the chart is a very highly reflective black. It's annoying in that manner. Wish it was matte; the only way to really shoot that chart without the reflections is to hide behind a big black sheet of duvetyne...

In clip 000 are you inserting the ND filter? and closing/opening the iris ? the shadow of the monsters hand across the girl in pink ( who now does this indian sitcom) looked like a hair in the lens...

No, I held up a sheet of diffusion in front of the light. I blared a hard open-face 500w light onto the scene for a couple of seconds, then I held up diffusion to soften it out, then brought it back out to harden it up again at the end, so we could hopefully get a variety of textures in the shot.

no noise for me so far, will check FCP transcode later

Very interesting, yes please let us know how the transcode goes.

 

[Barry_Green]

ZERO NOISE....PERIOD. As I have observed in all of the other AF100 footy. What ISO here Barry? (For documentation purposes only here)

"normal" 400 ISO. Interestingly enough, this is the exact same camera that was used to shoot Leo's footage, and the exact same camera that was the "A" camera on "A Verse Before Dying".

 

[Barry_Green]

I still think it has a lot to do with people's monitors. The stock settings for the Nvidia cards have the gamma way too high, bring it down and much of the "noise" goes away. My laptop has settings for overall computer and video only.

I agree totally. Between the stock computer gamma being too high, and then the Apple gamma shift bug hitting some folks, I think there's a lot to that. The Windows 7 calibration tool results in bringing the gamma down.

But I think there may be more to it. That's why I'm asking folks to do both -- calibrate your monitors, and report on your hardware/software combination.

 

[Barry_Green]

BUT... any idea how we can get the Panasonic engineers involved? Has Jan expressed to them these issues? I know she has said the problem is in post so are they addressing the problem?

I have asked them to conduct their own research, but I don't know that waiting for that will be the best use of our time, as who knows how long it might take them? I had this idea in the middle of the night and wanted to jump on it because I'm 88 to 94% certain that we're on the right track, and stuff like the QT7 vs. QT10 reports and the VLC reports are just confirming it for me.

 

[Razz16mm]

The .mts test clips played in Premiere Elements on my Toshiba laptop very clean. No visible noise. Vista 64bit core 2 duo with ATI Radeon 3470 gpu.

How does one get Quicktime player to play these?

 

[Kit Hannah]

I agree totally. Between the stock computer gamma being too high, and then the Apple gamma shift bug hitting some folks, I think there's a lot to that. The Windows 7 calibration tool results in bringing the gamma down.

But I think there may be more to it. That's why I'm asking folks to do both -- calibrate your monitors, and report on your hardware/software combination.

I brought the gamma down by about 10% and the noise from the candle video all but disappeared (very minimal noise in blacks). I will attempt to do another proper calibration on my laptop and see if it changes anything.

 

[Rick Burnett]

I still think it has a lot to do with people's monitors. The stock settings for the Nvidia cards have the gamma way too high, bring it down and much of the "noise" goes away. My laptop has settings for overall computer and video only.

People's monitor are different, but I think you've missed all the discussion of where this came from. People are comparing OTHER footage as well outside the AF100 on the same monitors and not seeing this level of noise at the lower gamma settings (the higher the gamma setting, the better it looks on mine) which is what sparked the questions to begin with.

The noise looks MUCH better on these. Notes:

1) NO noise on the color bars in this download or before.

2) On the still frame, I see noise in the gray rendition of the blacks on the edge of the shrink wrapped DVD. EVERYWHERE else is perfect. It's not horrible, but if this were ISO400, I'd be surprised to see so much noise in that area.

3) The test chart shows the noise in the last two gray bars only and no where else. It's not dancing though.

 

[Steve Kahn]

I had this idea in the middle of the night and wanted to jump on it because I'm 88 to 94% certain that we're on the right track, and stuff like the QT7 vs. QT10 reports and the VLC reports are just confirming it for me.

I think you are correct. This issue does seem to be multi-factorial, that is it seems to be both a gamma calibration AND AVCHD interpretation issue we're having. That would explain why the A cam on Verse can look good and horrible to my eyes.

Interpret AVCHD on the "right" computer and you're golden.

Poor gamma calibration can wreck any footage. Take the F3 footage Bloom just shot with tons of blacks. Throw off your computer gamma and it looks terrible. But, short of crushing blacks, I cannot completely "fix" the Leo footage for the life of me no matter how I fuss with the gamma. It is still there.

 

[Kit Hannah]

People's monitor are different, but I think you've missed all the discussion of where this came from. People are comparing OTHER footage as well outside the AF100 on the same monitors and not seeing this level of noise at the lower gamma settings (the higher the gamma setting, the better it looks on mine) which is what sparked the questions to begin with.

Nope, I didn't miss anything. Re-calibrated my monitor (again), footage looks great to me. Other footage looks great. Footage I have shot with my 5D, HPX500's, GY-HD110's, etc. All look great. I dunno...

 

[Rick Burnett]

Nope, I didn't miss anything. Re-calibrated my monitor (again), footage looks great to me. Other footage looks great. Footage I have shot with my 5D, HPX500's, GY-HD110's, etc. All look great. I dunno...

Well I watch footage on my MacBook pro from Red One MX, HVX200, 7D, 5D, GH1(3), GH2 and the various clips from the AF100 that have been made available. If I am not mistaken it seems a lot of MacBook Pro people are seeing similar results? (is that true?)

I can adjust the gamma on every clip except the candlelight clip to make the noise go away, but that doesn't explain the difference in noise compared to other footage I have shot with other cameras/codecs. I am not saying what the problem is, just that there is some difference that some of us are seeing is SOME of the footage.

Even with my MacBook calibrated and the gamma set at 2.2 I am seeing more noise in the ISO400 shot on the DVD edge than I would have expected. In a few days, I can at least compare the 7D and AF100 on a similar setup to try and help. EVERYTHING else in that shot looked great and the black around the picture frame looked perfect. I don't understand why that one area of gray is so pronounced on my system.

Barry, do you have a nanoflash there? Can you record the same scene internally and to the nanoflash and upload for me to look at on this monitor? I can at least tell you if they appear different to me. At least from this end I can tell you how different the noise is on a still shot.