HuffyUV question.

Coderjo · Post by **Coderjo** » Mon Apr 24, 2006 2:43 pm

Keeper of Hellfire wrote:I've read somewhere (sorry, don't remember where) that it can be a good idea to enable RGBA even if an alpha channel isn't used. The reason for this sounds plausible: Having 32bit per bixel allows faster computing than 24 bit per pixel since the CPU works with 32 bit (or multiples of them). But I'd never checked if it's true.

Actually, there is merit to this claim. CPUs tend to operate better with data aligned on a multiple-of-four boundary. RGBA, being 4 bytes, would keep the alignment across every pixel in the image. You do give up a bit more on compressability if you use this, though, since you are feeding 25% more data to the compressor.

RamonesFan2020204 · Post by **RamonesFan2020204** » Mon Apr 24, 2006 8:46 pm

I wish there was some way that I could watch both versions of the vid I made side to side to see the difference.

RamonesFan2020204 · Post by **RamonesFan2020204** » Mon Apr 24, 2006 8:46 pm

And by side to side, I meant two or more versions at once.

Coderjo · Post by **Coderjo** » Mon Apr 24, 2006 9:20 pm

you can, using AVISynth. Load the videos into variables and then use StackHorizontal or StackVertical.

RamonesFan2020204 · Post by **RamonesFan2020204** » Mon Apr 24, 2006 9:27 pm

Coderjoe wrote:you can, using AVISynth. Load the videos into variables and then use StackHorizontal or StackVertical.

What's that mean? And did you get my PM Coderjoe?

trythil · Post by **trythil** » Mon Apr 24, 2006 9:37 pm

RamonesFan2020204 wrote:
Coderjoe wrote:you can, using AVISynth. Load the videos into variables and then use StackHorizontal or StackVertical.
What's that mean? And did you get my PM Coderjoe?

They mean what the function names say they mean: place clips side-by-side horizontally or vertically.

Code: Select all

a = [clip1]
b = [clip2]

StackHorizontal(a, b) / StackVertical(a, b)

Obviously, you should replace clip1 and clip2 with appropriate source functions -- go read the Avisynth manual if you don't understand that. And you should only pick one of the Stack* functions.

Keeper of Hellfire · Post by **Keeper of Hellfire** » Tue Apr 25, 2006 1:54 am

RamonesFan2020204 wrote:I wish there was some way that I could watch both versions of the vid I made side to side to see the difference.

If you have installed AviSynth from the AMVApps then you have the function BeforeAfter(...), which was made exactly for that purpose - to see how a filter chain changes your source.

RamonesFan2020204 · Post by **RamonesFan2020204** » Tue Apr 25, 2006 1:58 pm

Okay, what happens when you check "Enable Full Size Output Buffer" and "Swap Fields on Decompres"?

DJ_Izumi · Post by **DJ_Izumi** » Tue Apr 25, 2006 2:34 pm

RTFM. :/

http://neuron2.net/www.math.berkeley.ed ... onfiguring

Sir_Lagsalot · Post by **Sir_Lagsalot** » Tue Apr 25, 2006 4:26 pm

Coderjoe wrote:
Keeper of Hellfire wrote:I've read somewhere (sorry, don't remember where) that it can be a good idea to enable RGBA even if an alpha channel isn't used. The reason for this sounds plausible: Having 32bit per bixel allows faster computing than 24 bit per pixel since the CPU works with 32 bit (or multiples of them). But I'd never checked if it's true.
Actually, there is merit to this claim. CPUs tend to operate better with data aligned on a multiple-of-four boundary. RGBA, being 4 bytes, would keep the alignment across every pixel in the image. You do give up a bit more on compressability if you use this, though, since you are feeding 25% more data to the compressor.

Enabling RGBA will hurt performance in any decent codec since 25% more data needs to be compressed. If aligning pixels to 32 bits is faster, then the codec should automatically convert video that isn't aligned; and whether or not the RGBA option is selected should have no affect on whether or not the calling program feeds it 32bit video.