Gamma

[JarrettH]

I've noticed with different matrices the gamma is noticeably darker. For instance SoulHunter V3, and believe me, this isn't the only one, looks darker than HVS-Good. When the gamma is higher in HVS it can make it look worse because it exposes more. I could raise the gamma, but I won't be able to make it exactly like the other picture. Is there reason for this? Does it have to do with HVS and MPEG compression?


Thanks;)

[Didée]

Please post the *exact* procedure you used to compare the clips to find out that effect you're describing.

[JarrettH]

Using BSPlayer, I open each clip in a different session (so they are both open at the same time). I can clearly see that the gamma is lower on one than the other. I skip to the same frame. Not much else to say.

[communist]

:p
Open the same clip twice in 2 BSPlayer sessions - woop-de-doo the gamma is different.
This has to do with overlay / rendering not with the actual encoding of certain gamma values. A (custom) matrix shouldn / cant change gamma of an encode.

[Didée]

Originally posted by JarrettH
I open each clip in a different session (so they are both open at the same time).
That's exactly what I wanted to hear ;)
Originally posted by JarrettH
Not much else to say.
Indeed: Not much else to say.

When you open two clips in two instances of a DS player, then the first player will use hardware overlay to display the video. For the 2nd one, no hardware overly surface is left, and so this one will be shown through DirectDraw software overlay. And the colors shown by HW overlay are always different (and aways darker) than DD mode. That's all.

If you want to compare side-by-side, either open your clips through Vdub, or open a dummy clip in your player, minimize & forget it, and THEN open the clips to compare in different player instances.

Gamma difference through different matrices, hihi ...

[edit]ah, I was too slow.

[JarrettH]

:p

Thanks, that's interesting. I didn't know that happened.

[JarrettH]

I just opened them independently using the same player and the results are still the same. It's no big deal really.:)

[Didée]

I don't know what's going wrong, JarretH. But it is definetly going wrong on your side.
Originally posted by JarrettH
I just opened them independently using the same player
Perhaps something fishy is happening on decoding stage (BTW, what player?). Try opening them VirtualDub, and look again. If you have AviSynth installed, try the following:

source1 = AviSource("firstvideofile.avi")
Source2 = AviSource("secondvideofile.avi")

stackvertical(source1.colorYUV(analyze=true),source2.coloryuv(analyze=true))

Open that in VirtualDub, and look at the numbers. The *differences* for the average values will be in the range 0.00 ~ <<1.00.
If the difference on the Y channel is constantly bigger than 1.0, one could see a difference, if it's smaller, then it's the normal result of different quantization.

From the other thread (http://forum.doom9.org/showthread.php?s=&postid=547193#post547193):
He convinced me.:p
You didn't realize that you posted this (in referal to the thread starter, I assume) directly after a post by syskin, who stated that it IS NOT SO. Let me tell you that syskin is an XviD developper that knows the codec like his trouser pockets. If he says "XviD doesn't do that", that means that XviD does not do that.

It is simply impossible from a technical point of view. Your car won't run slower or faster, depending on the jacket you're wearing.

[Soulhunter]

Originally posted by Didée

Your car won't run slower or faster, depending on the jacket you're wearing.

I could swear my car runs slower if I wear my 1000 pound lead jacket... :D :D :D


Bye

[JimiK]

You got a 1000 pound lead jacket? Interesting... :)
@JarrettH Did you try just to open the same file twice and see wether you still get a different gamma or not?

[Soulhunter]

Originally posted by JimiK

You got a 1000 pound lead jacket? Interesting... :)

Yep, since my hollidays in Tschenobyl... :D


Bye

[JarrettH]

lol, pictures;)

It seems a wasted effort to reply like that if you don't "know" him (syskin).

I'll try them in VirtualDub then. Are there any settings in the decoder I should be using or not using that could be affecting it?

[*.mp4 guy]

:confused: :o ... eheheh I guess I was very wrong about that sorry. I thought I read an article somewere that divx did something like that if you recompressed its own output several times at a quant of 1 anyway sorry to be spreading misinformation it was a stupid mistake.

[JarrettH]

On a side...

So my system has become very unstable as of late, won't get into details, but after deleting about 30gb of stuff mostly games, and other space hogging nonsense, I went to encode Lost In Translation using SixOfNine=20 - the last one I wanted to try. First pass is usually 1hr 20-30min, now it estimates 22min!!:eek: Amazing.
Everyone clear your clutter! My system won't be in this barren state of course once I format and use SP2, but it is interesting what an enormous affect having large amounts of files does. I could recommend to put VirtualDub on a separate partition or have it as one of the first things you install.:)

[JarrettH]

Meh this can't be right...I'll see the results in the morning.:cool:

[JarrettH]

Never mind, it seems either DVD2AVI or VFAPI need to be in the same location. Can't just run open the VFAPI file with VirtualDub and expect it to work properly.

[Koepi]

VFAPI?

VFAPI converts colour spaces and thus can affect gamma (no, not depending on quant matrices used).

You may want to get on the "pure YV12 processing chain"-track, read the avisynth forums/stickies for more details. (In short: it's faster, it's more precise due to no colour space conversions...)

Thank you for providing us with the necessary information ;)

Regards
Koepi