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

Yep, I can definitely confirm that you can make the camera as clean or as noisy as you want. :) I will also say that the H.264 also smooths out noise as well, which I've seen when comparing internal footage to external recorded footage.
 
Well I've been following the noise thread(s), have made adjustments in settings, ISO, etc. but have definitely not solved an intermittent issue with noise in the midrange (i.e. the faces) in interviews. Literally seems to come and go within clips. Disturbing.
Usually happens when my background is dark (purposely) and the head has the highest IRE value (around 70 at brightest points on face). In this Vimeo clip it's very hard to see, sorry, but it's there not only on my computer monitor but on my TV Logic, SDI-d from camera during shooting so this is not a computer monitor issue. I used the Abel Cine Clean settings recently which definitely has helped with noise in dark areas. But not in the midrange, so if anyone has experienced same please reply.
Again seems to be a problem only when a lot of the frame is dark, but as I say, I would like to have the option to shoot this way anxiety free.

Clip is:
http://vimeo.com/33083894

Thanks
 
Those interviews look great -- I didn't see a speck of noise in it. If you're worried about noise in faces, have you looked into using the SKIN TONE DTL setting? It makes a rather substantial difference.
 
It may not be noise in this case.

I agree with Barry but add...I think that sometimes HD gives us a kind of "shock and surprise" in portrait and face work... because of the level of detai.

Skin has pores and is alive and moving. It has detail and texture. It shines and has various tones. Its a dynamic moving target for the sensor.

This detail is not always flattering (even with make up).

I am now using a black promist 1/8th and a little post work for this situation.

I have never tried the Skin Tone Setting ( keep forgetting about it)...but I will test and try it one day.
 
Thanks to you both (Shooter / Barry). I know you can't see properly in the clip I posted. It's an artifact closest to dirty heads from back in the tape days, not like most of the noise discussed earlier in the thread. I'll keep on with this, play with the skin detail, and try the softening filters I still fortunately have.
Best, John / JDM
 
Well I've been following the noise thread(s), have made adjustments in settings, ISO, etc. but have definitely not solved an intermittent issue with noise in the midrange (i.e. the faces) in interviews. Literally seems to come and go within clips. Disturbing.
Usually happens when my background is dark (purposely) and the head has the highest IRE value (around 70 at brightest points on face). In this Vimeo clip it's very hard to see, sorry, but it's there not only on my computer monitor but on my TV Logic, SDI-d from camera during shooting so this is not a computer monitor issue. I used the Abel Cine Clean settings recently which definitely has helped with noise in dark areas. But not in the midrange, so if anyone has experienced same please reply.
Again seems to be a problem only when a lot of the frame is dark, but as I say, I would like to have the option to shoot this way anxiety free.

Clip is:
http://vimeo.com/33083894

Thanks


The fact that you are seeing the issue live rules out a post production issue (like long GOP issues, or codec issues)

I believe it's a lighting and sensor sampling frequency issue.

It's very subtle in that clip, but if you look at the man's forehead, there is a sort of 'undulation'. (not talking about when the camera shifts or exposure changes)

If you look at a grey scale clip, or exagerrate the luminance it will be more apparent, you will see "bands" flickering on the forehead, not unlike a fluoro flicker. Some people are very sensitive to these issues, and many do not perceive it at all.

If this isn't what you are talking about, please describe it more thoroughly or upload a clip that demonstrates the issue more clearly
 
Could it just be the rolling shutter/shutter speed/fluo light issue? JDM, what shutter speed were you shooting at, and under what kind of lighting?
 
Barry, if you record 422 out of the HDSDI on the AF100...shouldn't then (in theory based on what you're arguing for in this thread) the footage look much cleaner? I've read the difference not being that great between external footage and internal in terms of how it "looks."
 
That's because the onboard codec is a lot better than people give it credit for. It's quite good.

SDI footage will be the uncompressed 4:2:2 footage, as best the camera can make it. That will then be compressed and recorded by an external recorder, which will close the gap somewhat between the onboard codec and uncompressed; the better the codec you use, the more of the original goodness will be retained.

If AVCHD was a terrible codec, yes you'd see a huge difference between the onboard and the external. But because AVCHD is surprisingly good, you don't usually see that much of a difference at all. Which is why I've been telling folks for a year now -- don't sweat an external recorder, until you've given the onboard a good thorough examination, because the onboard may very well be good enough. And if it isn't good enough, then sure, go with an external, and gain the benefits of 4:2:2 and simultaneous recording for redundancy purposes. That's what I use the Nano for. But I'll also be the first to tell you that you have to usually look long and hard to see ANY visual difference between what the Nano records, and the onboard codec.

Okay, last thing -- an external recorder might show more noise than onboard recording, because noise is exactly the type of thing that a codec will toss out first. The more robust or uncompressed the codec, the more noise you'll see.
 
I will absolutely confirm just what Barry said.

I have just completed a mornings worth of test recording simultaneouslyto both internal AVCHD and the Samurai (Prores HQ / Prores 422) .

I have been stress testing and pixel peeping it in my Adobe CS5.5 suite.

I place one clip on one video layer and the other above it ...then flick between them.

I have applied exact same filters with exact same settings such as CC. , Curves, RGB Correction, Colorista, Levels, Leave Color, Chroma Key etc to each. I can see or detect no difference in them. Occasionaly a subtlety can be observed but you have to look close and hard.

Now what becomes more interesting ...is when one looks at the scopes as one flicks between the two types of the same source clip. THere are some observable differences but they are largely beyond my technical expertise to interpret with any certainty.

I would state that there definitely is something that goes on in the luma level of the signals on the scopes. The Prores actually clips the highlight end of the signal and the AVCHD lets it go.

The Prores appears "cleaner" in the Premiere scopes. ie. the generated bars signal (ex camera) is devoid of noise in the scope.

There is alway a degree of shift in the scopes when flicking between the clips on live and generated footage.

Now... at this point I have no idea what part Premiere is playing in all this.

I need to see it in a different post suite next.

Meantime...the AVCHD codec is not to be dis-respected at all. ( I have generally alway felt that about it from my results and experiences ...but now I know a bit more since testing for myself)

BTW - I tested "motion" (camera movement visual effect on a subject). I thought and hoped that there maybe an advantage with a higher data-rate codec. I cannot see an advantage in my suite but I will with hold opinion until I see it in a Flame Suite.
 
I know they're comparable, I was talking about the likelihood of seeing noise in the external as opposed to the internal since the computer wouldn't be decompressing h264 with the external like it would need to do with the internal footage. That's why I was asking. I suppose if I convert the internal footage to Prores the result I see on the computer should be similar to externally recorded Prores right?...in terms of how my computer displays the noise.
 
I know they're comparable, I was talking about the likelihood of seeing noise in the external as opposed to the internal since the computer wouldn't be decompressing h264 with the external like it would need to do with the internal footage. That's why I was asking. I suppose if I convert the internal footage to Prores the result I see on the computer should be similar to externally recorded Prores right?...in terms of how my computer displays the noise.
Not necessarily -- once again, you're bringing the computer into the equation as a a variable, and so it all depends on how your computer decodes the footage. Look earlier in the thread, we had reports of people saying VLC looked a lot noisier than other playback methods, for example.

My argument is that the actual footage recorded is clean as a whistle and good to go. And that those who were seeing aberrational noise were likely doing so because they were either viewing it on a monitor that was enhancing the noise (a six-bit panel or super-sharpened detail or whatever) or that their h.264 decoder was introducing noisy playback. We've seen that behavior in the P2 Viewer decoder, and we've seen it in VLC.
 
yeah, well said, less compressed footage is always noisier because there are less blocky sections due to compression and instead we get pixel accuracy, which is noise. So meh, i like noise. Prefer noise over blocks :D
 
I think my Dynamic contrast enhancement in the Nvidia video settings control panel induced A LOT of noise.

I set the gamma in the windows color manager and turned that Nvidia nonsense of, now I have to look for noise and blockyness :)
 
Last edited:
Hey all, been going over this thread (sooo long) and have noticed some really interesting information. I'm the same in that my footage on a hd tv often looks amazing, but on my laptop it looks noisy etc. (But then when i play blu rays they also have the same apparent issue). To be a tiny bit lazy (it is 32 pages of forum stuff to sift through) I was wondering if someone could sum up what you guys and barry concluded? Was there a way to fix this noisy codec issue?

I was debating on getting an eizo fs2331 monitor but now am put off, would connecting an external monitor for preview/noise check purposes still show the same noisyish issues i have on my laptop? Would I be better off getting a hd tv? (either way i wil calibrate with a spyder)
 
...my footage on a hd tv often looks amazing, but on my laptop it looks noisy etc. (But then when i play blu rays they also have the same apparent issue). To be a tiny bit lazy (it is 32 pages of forum stuff to sift through) I was wondering if someone could sum up what you guys and barry concluded?
These problems are all symptoms of the infamous Quicktime H.264 "gamma bug". Here's a link an article that explains the technical issues in merciless detail:

http://vitrolite.wordpress.com/2010/12/31/quicktime_gamma_bug/

Adobe fixed the CS5 encoding flaw described in that article in CS5.5 and CS6. The H.264 decoding and display glitches are hazards of using Quicktime-based video players. Here's a link to a recent discussion on that issue:

http://forums.creativecow.net/thread/20/868723
 
hey thanks for the fast responses, had a quick look at the links and all look very helpful, they seem to talk more about colour though while im more bugged about the noise grainyness issues. But will read into them more later. Bit worried too as i already have cs5.5, and you say it that version they fixed this? I'll read more into it later this evening, what do you think about going for the eizo monitor or should i be focusing on a hdtv for preview purposes to avoid this noise?
 
hmm just looked into it a little more, they're talking about the h.264 codec, but even my raw avchd footage i copy over from my camera looks noisy etc! So perhaps this h.264 gamme shift is not the issue for me :/
 
Back
Top