Brightness when making DVD from DV camera using Premiere

[andie41]

trevlac,

As far as the WFM. Break it out and test your DVD player and CAM. I'd bet the cam outputs IRE0 and the DVD player does 7.5. Only way to know is in the Analog world.

You are right; I have been putting off trying to set up the WFM because it is going to be a pain and my expectations are, after several years in a box, it is not going to work anyway, but after all the fiddlin' I have done ...

Thanks for the suggestion Re not checking the infamous "output YUV ..." box - that is certainly my conclusion, as well. This has become urban legend. There are many comments and questions around the net about this brightness issue and it has become amusing, as one person will say "definitely check the "output YUV ..." box, and in another thread the response is to "make certain you do not check the "output YUV ...." box. The only reason I can see to check this box would be if you were only watching your video on the PC, because it most assuredly expands the black level to illegal NTSC values. In one test, I had a contrasty scene (indoors with sunlight streaming through the window) and there was a large block of pixels in the 16-40 (approx) range (when looking at the histogram in VDub) and when this was encoded with the YUV added, this range was now 0-20 - almost all clipped RE TV viewing.

I have altered my method of examinng my various video examples. Rather than copy a still and paste into Photoshop, I look at the histogram in VDub and this is totally consistent with my previous results. For whatever it is worth, there is an option to view in luma rather than RGB. I select this, although I can't see any difference. Also I crop 8 pixels all around and change the crop color to middle grey so as to eliminate the black border from influencing the blacks in the histogram.

I have become more impressed with TMPGenc, after these past few days (other encoders may do the same). The program appears to do quite a good job of adjusting luma values. from a DV file, but not over doing it. It seems to expand, if necessary, the black levels down to around 8 (with the bulk of the black pixels beginning in the 16 range). As I stated before, when TMPGenc receives uncompressed RGB (either fron am uncompressed AVI or from VDub, it leaves the values alone. I like that. If I want to manually adjust my video, I can send it to TMPGenc and it won't mess with it.

I have a sony camera and hence my DV's are encoded with the Sony codec. Generally I use the Sony codec on my computer. I decided to try a different codec (Mainconcept) for the decoder on my computer. What a difference. In the few examples I looked at, MC appears to expand the luma out and make a much flatter histogram. This doesn't seem necessarily desirable., as in the contrasty scene I mentioned above. The Sony histogram showed a more uneven histogram, to me, indicative of the scene. However, regardless of which codec was installed when producing mpegs, TMPGenc produced what appears to be an identical histogram - so it is doing it's own thing.

So my question and my concern is on what basis do these programs and codecs adjust the values. Are we attempting to make a silk purse from the sow's ear? - the data is either present or in some way manufactured. The Sony codec shows a particular histogram and MC has another. As this was encoded with the Sony, where does MC get this other data. And when TMPGenc gets through with it, it will be different again.

[trevlac]

#1
the infamous "output YUV ..." box -
....
The only reason I can see to check this box would be if you were only watching your video on the PC, because it most assuredly expands the black level to illegal NTSC values.

#2
I look at the histogram in VDub

#3
I have a sony camera and hence my DV's are encoded with the Sony codec.

#4
So my question and my concern is on what basis do these programs and codecs adjust the values.

1. There are very good reasons for using that check box. For example, one of the better flows for analog capture is: 1) capture YUY2 in mjpeg 2) when mjpeg outputs RGB it does not change the nominal range. So RGB from YUY2 comes out as 16-235. This is a minimal effect way to go from YCbCr 4:2:2 to RGB. 3) Use the box in TMPGEnc and it will convert from RGB to YCbCr 4:2:0 again without messing with the range.
   
   The range change is effectivly like changing the brightness and contrast to meet a standard.
   
   
2. I'm confused about this. I don't use vdub for capture, but isn't that the only way you can see histograms in vdub? Here is a filter that does histograms, WFM, and a vectorscope soon.
   
   
3. But are you sure that the Sony code is DEcoding the file? Can you test the DV options I pointed out in TMPGEnc? I'd be real interested to know. Maybe I should setup a test myself? I don't use TMPGEnc...but.
   
   
4. The reason is to change from PC RGB standard to digital video YCbCr standard. Unfortunately, the methods they use can give different results (formulas, rounding, interpolation,...).
   
   I see 2 reasonable approaches to dealing with this. 1) Never convert like FulciLives said. 2) Convert asap with something that works the way you want. Here, the conversion back by TMPGEnc is still possibly unknow (but testable).

[andie41]

As I was saying, one should always select the "out YUV as basic YCbCr not CCIR601". I didn't say that? Well I meant to say that. Before the aumbulance arrives to take me to the padded room, I thought I would report my latest findings.

I did get my WFM setup.Did I hear someone say, "ignorance is bliss?" According to the WFM, the colorbar pattern shows the following from my DVD player. These were from mpegs made directly in TMPGenc.

Uncompressed avi without the "output YUV" 10 IRE - 90 IRE
Uncompressed avi with the "output YUV" 5 - 98 (approx - slightly below the 7.5 and 100 lines)
Sony DV AVI without the "output YUV" 5 - 98
Sony DV AVI with the "output YUV" 5 - 98

Results from mpegs frameserved from VDUB without output YUV selected were 10 - 90 and when output YUV was selected, 5 - 98.

Output from the computer NLE timeline through the Sony camera is always 0 -90. Is the WFM accurate? I tend to think so. One (of many) thing that confused me, was that the pluge pattern disappears from the mpeg using the "output YUV". I was looking at this and concluding that TMPGenc was converting all these black and grey vaues to a level too low, but perhaps it was just losing the dark greys in the conversion.

Also the chroma values were almost right on in the Vectorscope from the Mpeg's that had been encoded with the "output YUV" selected.

Maybe my monitor is just off. Maybe the operator is off.

It must be 5 O"clock Somewhere ...