What's the difference between slow ist pass and fast 1st pass?

[PiPPoNe92]

Which the difference beetween slow-firstpass and fast 1st pass?
I know that fast first pass is:
* ref 1
* no-8x8dct
* partitions i4x4 (if originally enabled, else none)
* me dia
* subme Min(2,subme) (if original value is greater than 2, change it to 2, else do nothing)
* trellis 0

but, I want know how much quality I gain if I'll use slow-firstpass than fast 1st pass.

while we're, I want know how much quality I gain if I'll use me: tesa instead of me:umh (if possible the value will in percentage).

Thanks for your answers..

[LoRd_MuldeR]

Which the difference beetween slow-firstpass and fast 1st pass?
I know that fast first pass is:
* ref 1
* no-8x8dct
* partitions i4x4 (if originally enabled, else none)
* me dia
* subme Min(2,subme) (if original value is greater than 2, change it to 2, else do nothing)
* trellis 0

but, I want know how much quality I gain if I'll use slow-firstpass than fast 1st pass.

while we're, I want know how much quality I gain if I'll use me: tesa instead of me:umh (if possible the value will in percentage).

Thanks for your answers..

The "fast first-pass" option, which is enabled by default for a reason, only lowers settings in the first pass that have a very minor effect on the final video from the second pass.
So you probably wouldn't be able to tell the difference between the video encoded with "--slow-firstpass" and the one encoded without that option. But the later certainly will save a lot of time!

Also the fact that "--me tesa" is only used in the "placebo" preset (it's not even used in the "veryslow" preset!), should tell you what to expect from that option ;)

[PiPPoNe92]

you don't answer to my question yet.

[buzzqw]

try yourself and judge (and post psnr/ssim results.. and also screenshots)

btw.. maybe on your monitor (or any users monitor) you cannot see all the difference on slow/fast...

BHH

[LoRd_MuldeR]

you don't already answer to my question.

What do you want to hear? I think the answer was quite clear. But again:

Both options, "--me tesa" and "--slow-firstpass" will be significantly slower, but the quality improvements usually will be very small (if visible at all).

I certainly wouldn't recommend to use such "placebo" options, unless you have a lot of time to waste...

try yourself and judge (and post psnr/ssim results.. and also screenshots)

Metrics like PSNR or SSIM certainly are not a good way to judge the overall visual quality. One should rely on its eyes for that purpose!

Also, if you really want to look at metrics, you should use the "--tune psnr" or "--tune ssim" option! Otherwise the Psy optimizations in x264 will screw up your PSNR/SSIM values :eek:

Needless to say that without those Psy optimizations the visual result probably will be much worse...

[PiPPoNe92]

I use --tune film option, because I rip only films.

[LoRd_MuldeR]

I use --tune film option, because I rip only films.

That's generally a good idea. But it wouldn't be a good ideo to use "--tune film" and then look at PSNR or SSIM values!

If you are interested in metrics, use the corresponding "--tune" option. But again: Metrics are not a good way to judge the overall subjective quality.

You should better make two 2-Pass encodes of the same size (same target bitrate) and compare them visually...

[buzzqw]

Metrics like PSNR or SSIM certainly are not a good way to judge the overall visual quality. One should rely on its eyes for that purpose!

that's why i ask for screenshoot too

since OP ask about a % of gain, a numerical value psnr/ssim could give numbers to compare ;)

BHH

[LoRd_MuldeR]

since OP ask about a % of gain, a numerical value psnr/ssim could give numbers to compare ;)

Numbers are nice, but not very helpful in this case. Hows does "1% better" look in reality ???

Also a single screenshot doesn't tell the whole truth. He should look at the video while moving and then decide.

Still you can assume that the x264 defaults and preset names have been chosen for a good reason ;)

[buzzqw]

Also a single screenshot doesn't tell the whole truth. He should look at the video while moving and then decide.

that's a huge work works for PiPPoNe92 :D

Still you can assume that the x264 defaults and preset names have been chosen for a good reason

that's why my presets on HDC are very easy to use :)

VGG