Survey: What min/max levels does your cam shoot?

Back to post

Comments

farss wrote on 6/21/2011, 8:35 PM
Sorry not the definative test, hard for me to get my hands on a 5D MkII for very long so this is from actual footage supplied by someone a long time ago:



I thought I'd include it as there's a dearth of results from the DSLRs.

As this shows, the vDLSRs shoot Computer RGB, there's nothing in the original scene that's "absolute" black hence nothing in the histogram that goes down that low.

here's another one from my SN9C01 webcam, darn impossible to run the tests with this thing, the autoexposure / auto-illumination is very savage:



enought there to say this thing is also Computer RGB.

Bob.
john_dennis wrote on 6/21/2011, 9:07 PM
I frequently shoot video with my Canon G9 though I'm looking for something with higher resolution.

Black


I'll have to wait to do white until tomorrow. My efforts to shoot a white sheet of paper with any light I have tonight has given me poor results. I repost tomorrow.

White
NickHope wrote on 6/21/2011, 9:32 PM
John D, why not just shoot the light itself? Max out your exposure and zoom in on the lightbulb itself.

Thanks for the feedback everyone. I have updated the bottom of the first post with the results, and will keep it up to date with any new ones.

Most interesting one for me so far is Fuchs' Panasonic Lumix GH2. A dSLR that is shooting nominally 16-235 and not 0-255.
musicvid10 wrote on 6/21/2011, 10:34 PM
Testing a couple of older camcorders, I discovered that the stock lens caps are neither light-tight nor completely opaque.

Recommendation for the lenscap test: As Nick described, and inside a darkened closet with the door closed.
farss wrote on 6/22/2011, 1:21 AM
Nick said:
"Most interesting one for me so far is Fuchs' Panasonic Lumix GH2. A dSLR that is shooting nominally 16-235 and not 0-255. "

This is where it gets interesting or is it confusing. As Fuchs said:
" In real-life recording you will often have peaks below RGB 16 and above RGB 235 and in one or two shots out of hundred shots it may reach from 1-254."

From my experience a camera that goes more than a tiny fraction below 16 is not recording to Rec 709.

Hm, just downloaded the source video for this clip on Vimeo :http://www.vimeo.com/19276461

Here's the results at 21:09



I feel pretty safe in saying that camera is recording 0 to 255. It also appears to use "dynamic contrast control" which might explain the variable results, don't know for sure of course.

Bob.
Marco. wrote on 6/22/2011, 2:38 AM
I think we shouldn't guess on shots we don't know how the camera was set to when being used.
The GH2 is said to record sRGB when you set the preference to sRGB and leave i.dynamic off (what the standard preferences are).

All I can say is I never ever had a consumer camera or footage from a consumer camera (and I edited a LOT of footage from wide variing camera models) which did record such close to the BT.601/BT.709 when it's about the luma range. When exceeding the 16-235 range it's always been rare signals which aren't relevant (like sun reflections) and which you would usually just clip by a broadcast limiter while color correcting (because you won't notice). And this is what the under- and overexposer buffer of BT.601 and BT.709 is meant for. It gives some extra room to let as much of the relevant signals inside 16-235 as possible. I don't take that in a too technical manner. For me the GH2 just works like I would expect when matching BT.709. It works great for color correcting and grading high compressed 8 bit 4:2:0 signals.

Btw: I find the histogram bit misleading sometimes when it's used without some closer looks onto the waveform monitor. If you proof a wide range of GH2 shots via the waveform monitor it is rather clear it strongly tries to keep as much signals as possible between 16 and 235 and only rare - hard to see - oscillations come beyond.

Edit:
Just compared my GH2 footage with a bunch of XDCAM footage out of our broadcasting house, variing from IMX 50, 720p50 to 1080i25. They all behave the same way in terms of exceeding 16-235. It's just the buffer the BTs give us.
farss wrote on 6/22/2011, 3:26 AM
Fuchs said
" I find the histogram bit misleading sometimes when it's used without some closer looks onto the waveform monitor. If you proof a wide range of GH2 shots via the waveform monitor it is rather clear it strongly tries to keep as much signals as possible between 16 and 235 and only rare - hard to see - oscillations come beyond."

You are ever so correct there and thank you for saying that.
In fact the frame I used previously shows EXACTLY what you said to be true. Looking at the waveform monitor, set to 16-235 it shows completely compliant Rec 709 as shown below.



Personally I have never relied on the histogram, in my camera or in Vegas. The standard instruments in the video world are the waveform monitor and vectorscope.

All of which does beg the question, why are we suddenly using a histogram to judge levels?

Bob.
NickHope wrote on 6/22/2011, 3:41 AM
Judging by that histogram it looks like the GH2 is trying to shoot blacks at around 16 and giving some undershoot. The graph left of the knee at about 18 resembles noise I get sometimes after compressing footage that was previously within 16-235.

First reaction is I don't think I would want to advise someone to do a C-RGB to S-RGB on that footage before uploading to YT etc., but I'm downloading it now to take a better look.
NickHope wrote on 6/22/2011, 3:50 AM
All of which does beg the question, why are we suddenly using a histogram to judge levels?

As discussed above, using the histogram eliminates the variable of the "Studio RGB (16 to 235)" video scopes setting, which affects the waveform but not the histogram. I could have instructed people to set that but if we got some unexpected results then we could never have been sure that they had it set correctly.
Marco. wrote on 6/22/2011, 3:57 AM
I use the histogram mainly to judge the gamma area which is harder to do via waveforms.

I could show some waveforms from GH2 shots later which also shows what I meant.
farss wrote on 6/22/2011, 4:23 AM
Sure there is that variable and it's a pity that the Vegas waveform monitor doesn't show what levels it is set to, that would eliminate that as a variable if included in a screenshot.

On the other hand though what Fuchs is saying is quite correct, the histogram can much more easily give the wrong impression when looking at unknown footage. Looking at that GH2 footage's waveform and peering very closely I can see the very tiny amounts of the image going below 16. I'd actually hazard a guess myself that they might be the result of some edge enhancement, you get the same thing with Vegas using the unsharpen mask. As you quite rightly say they are best ignored and left to get clipped.

Following on from Fuchs's revelation I went back and rechecked the 5D MKII footage, according to the waveform monitor my opinion doesn't change, the blacks are solidly below 16, at around 5 actually and the highlights hard to 255. This footage should most certainly have a cRGB to sRGB conversion applied before uploading to YT.

My personal preference for the waveform monitor is that by looking at a frame from a high contrast scene I can pretty quickly know what to do with the footage, I oftenly have to work with footage from unknown cameras.

This is the whole problem that I think Musicvid has been struggling with. How to give foolproof advice when you don't know what camera is being used, how it has been setup etc. I still think his idea is as good as it gets for those who don't want to do any of the hard yards.

Bob.
Marco. wrote on 6/22/2011, 4:34 AM
Here are some GH2 samples with waveform view beside:

Sample 1
Sample 2
Sample 3
Sample 4

In this case Vegas preview was set to half size just because of simplicity to have both windows in same size beside. Also the screenshots were resized later. This makes the preview looking bit blury what the shots actually are not!. But I ensured the waveform still shows what I meant in my post before.
Marco. wrote on 6/22/2011, 7:55 AM
Another one:

Dell Streak, 2011
MP4, 720p

Black peak: RGB 0
White peak: RGB 255
musicvid10 wrote on 6/22/2011, 9:03 AM
Histogram displays RGB levels, not IRE (as does Waveform) with setup options that can be misleading. Parade displays RGB, but not luminance.
Kimberly wrote on 6/22/2011, 11:38 AM
If I understand the question correctly, my Sony HDR-HC3's shoot from 0 all the way to 255.
farss wrote on 6/22/2011, 2:08 PM
Kimberly said:

"If I understand the question correctly, my Sony HDR-HC3's shoot from 0 all the way to 255"

The Sony HC3, HC5 and HC9 all record to Rec 709. There is a caveat with these cameras, they can also record using Sony's x.v.Colour. I'd suggest checking you haven't got that turned on.

Bob.
Kimberly wrote on 6/22/2011, 4:40 PM
@farss:

Where in the menu might this be buried? I have the VASST DVD Inside the SONY HC Series Camcorders. I remember seeing something about Sony's x.v.Colour for the HC-9 but I think it is not available for my HC-3. Not that I understand what it is anyway : ) but now I'll have to watch the DVD again. It's a good refresher.
farss wrote on 6/22/2011, 8:11 PM
@Kimberley

I've only got our HC5 here. Shouldn't be that much different to the HC3 though. It'll be under P-Menu>Menu>Camera Set.

If your camera is capable of X.V.Colour it'll probably say so on a sticker on the outside of the flip out LCD screen.

Some info on what X.V.Colour is all about here.

One interesting part is:

"The xvYCC color space permits YCC values that, while within the encoding range of YCC, have chroma values outside the range 16–240, or that correspond to negative RGB values, and hence would not have previously been valid. These are used to encode more saturated colors. For example, a cyan that lies outside the basic gamut of the primaries can be encoded as "green plus blue minus red"."

Be thankfull you're using Vegas if you do record with this on, other NLEs would clip those negative RGB values.

What I cannot verify for certain is if the HC3 was X.V.Colour capable or not. I could be sending you off in the wrong direction, sorry if I have. I'm far more certain that the HC3,5,9 normally do record Rec 709 and that means nothing below 16 and some errant highlights above 235.

Bob.
NickHope wrote on 6/25/2011, 7:13 AM
Anyone got a Canon 7D? A Go Pro? A Nikon vidSLR?
farss wrote on 6/25/2011, 7:24 AM
"Anyone got a Canon 7D? A Go Pro? A Nikon vidSLR? "

I can get a GoPro.

There's a potential issue with some of the DSLRs if I understand what a lot of people do with them. On the 5D / 7D many switch them into medium contrast to keep the levels within 16 to 235. If that's correct then the answers you get back depend on how the camera is setup.
The same to a lesser extent applies to all the CineAlta Sony cameras.

Bob.
NickHope wrote on 6/25/2011, 8:51 AM
Ah OK, I didn't know there is a switch. Well, that's OK as long as I know. I'll do a bit of research on that. This is all background so I can categorise cameras for a rewrite of this article, so as long as I can say something like "* - depends on whether medium contrast is set", then that's OK.

I expect GoPro will be 0-255 but you never know. The Panasonic surprised us.
john_dennis wrote on 6/25/2011, 11:03 AM
Went through some of the video from my son's Sony SDR-SR1 AVCHD camera and found evidence it records 0-255. (See next post) 255 was easy, I found blacks as low as 3 in actual video.

Histogram
NickHope wrote on 6/25/2011, 11:21 AM
Thanks John. In view of results from the other Sony and AVCHD cameras, I'd expect the Sony SDR-SR1 to record nominally 16-255, not 0-255, but maybe with some "undershoot" noise below 16. Any chance of a histogram of some footage with some deep black in it? Or even better with the exposure minimised (dark room, lens cap on, iris close, fastest exposure, zero gain).
john_dennis wrote on 6/25/2011, 11:30 AM
I'll get the camera and do the test properly.

You are correct. Anything below 16 appears to be spurious noise.

Dark closet pointed at black fleece.