Differnent speed and size same CRF settings

[Bernix]

First video slow time big size (11hrs 6GB)
Source
Format : AVC
Format/ : Advanced Video Codec
Format_Profile : High@L4.1
Format_Settings_CABAC/String : Yes
Format_Settings_RefFrames/String : 2 frames
Format_Settings_GOP : M=1, N=10
CodecID : 27
Duration/String : 1 h 44 min
Width : 1 920pixels
Height : 1 080pixels
Aspect ratio : 16:9
Frame rate : 23,976 (24000/1001) FPS
ColorSpace : YUV
ChromaSubsampling/String : 4:2:0
BitDepth/String : 8 bits
ScanType/String : Progressive

Second video fast small (5hrs 1GB)
Source
Format : AVC
Format/ : Advanced Video Codec
Format_Profile : High@L4.1
Format_Settings_CABAC/String : Yes
Format_Settings_RefFrames/String : 4 frames
CodecID : 27
Duration/String : 1 h 33 min
BitRate_Maximum/String : 35,0 Mb/s
Width : 1 920pixely
Height : 1 080pixely
Aspect ratio : 16:9
Frame rate : 23,976 (24000/1001) FPS
ColorSpace : YUV
ChromaSubsampling/String : 4:2:0
BitDepth/String : 8 bits
ScanType/String : Progressive

He encoded the files to H256 (x265) CRF20. I know it depend on contents. The problem is that the second video with same setting was done quick and resulting in small size and ugly result (colors etc). Friend asked me why is it. He done the second video with CRF 16 then it lasted 8 hrs and size was 3,5 GB and result was good.
Only difference in source I see is in the first video Format_Settings_GOP : M=1, N=10 and in second BitRate_Maximum/String: 35,0 Mb/s but i dont know why it should have impact on resulting video.

Bernix

[badyu17]

x265? It says AVC in your media info example. The question is still vague IMO, there's no x264 settings used visible. GOP stuff means B-frames are not used and the keyframe interval is very short, only 10 frames. x264 uses 250 by default if you don't set it specifically.

Second example with maximum bitrate is probably output from tsMuxer. Eveytime you use tsMuxer you will end up with that flag added because that app modifies the sps in your video. I know because I used tsMuxer in the past a lot.

[Bernix]

Thank you for your response.
I added in info, that this is sources videos. I ask him to send me final video info. But why this sources are processed with different speed and size and quality with same settings + CRF20 (HEVC) that is the question.

Bernix

[badyu17]

If his source is bad to begin with, output will be bad as well no matter what crf he used. As you said it depends on the content as well. Is he using some transcoder application with different presets?

[Bernix]

I write him to send me the settings. Both videos was "superHQ" quality. Lots of GB each. First quality like was (if I remember correctly bit grainy) But I think I get something. No b-frames and each 10 frames I frame it means "bigger data" to process. But sources videos has similar size on other hand.... When in second video b-frames less data to process. But as I said, the quality of both videos was told me was excellent. So b-frame doesn't hurts video. It can only answer to speed up a encoding bit. And I think he is using vidcoder constant frame rate, CPU encoding and as I mentioned earlier CRF 20, with same settings for both video.

Thank you again

[Bernix]

If his source is bad to begin with, output will be bad as well no matter what crf he used. As you said it depends on the content as well. Is he using some transcoder application with different presets?

If source is bad, so CRF 16 wouldn't help (as you mentioned). Still waiting for his settings. He is using only one presets. Because 16 CRF helps, he is in doubt about constant quality and that is why I asking too. How reliable is same Preset with same CRF. Note that source with more grain was encoded properly at 20. Source with less grain not excepted quality with same preset and CRF. To me it is almost transparent CRF 20 x265. I have bad eyes, for me is CRF 24 good enough

Bernix

[_Al_]

CRF setting means that lower quantizer gives result that is closer to original, so yes CRF 16 would not help, if video is much worse than the other and CRF 18 or 20 was used.

encoder does not know if video looks better or worse, it only tries to get closer to original, lowering CRF value

You CANNOT compare CRF for different type of encoded videos. Even settings would not help. There is nothing to compare. If you theoretically want to match them encoded, you'd need deliberately worsen that better one, using much higher CRF. Because going opposite way, lowering CRF for that worse video, not only you'd ballooned video size but there is a chance you would not match the quality anyway.

[Bernix]

Hi,
I dont know if I understand you correctly. I know, computer cant see or compare video quality, but are you telling me that CRF 20 HEVC is for excellent source high (worse) CRF ? And quality of source as I was ensured was excellent, so I except result should be excellent. It is compared if Iam not wrong to CRF 14-15 x264 afaik people encoded videos at CRF 18 (x264). I know it depends on many factors, like subme and ref frames and ME and so on... but I believe they are good enough because encoding 100 min full hd on Core i7 for 11 hours that means cca 14 000 fphr that is cca 4 fps (what other setting must it be (i believe very high setting) And in second case get speed twice faster with identical settings... I know smallest data rate was OUT (therefore bad quality(but why)) but same was IN as in first case. So he came down to 16 which we can say is equivalent to x264 +/- CRF 10! and output was good compared to source. What is supposed to be transparent CRF in x264? I know you are telling me that there is not transparent setting but +/- it should be.
I still have not his settings, but as soon as I get it i post it.
So subjective quality at CRF 20 HEVC was bad and at 16 O.k. It seems to me ridiculous. Who use encoding in x265 with CRF 16. Anybody?
Thank you for your response, my english is very poor so I maybe get your post wrong.

Bernix

[badyu17]

The thing is, people here can't comment on something they didn't see. Did you see result of the encoded video? Did your friend explain why he thinks the result is bad? What is wrong with colors?

[Bernix]

The thing is, people here can't comment on something they didn't see. Did you see result of the encoded video? Did your friend explain why he thinks the result is bad? What is wrong with colors?

I know and I fully understand. Just want to add as I know him, he would be satisfied with acceptable quality. He complained mainly about bad colors, therefore I though if there isn't any wrong conversion from source 8, 10, 12 bit or 444, 422 or 420 or BT.601, BT.709 etc... But apparently to source this was not the problem because both file are in this way identical.
Thank you
Bernix

[Bernix]

Hi,
I get his settings soon. He is now unreachable for serious reason.

Bernix

[Bernix]

Finaly I got his settings. I hope they are correct. I personally disagree with using amp, because decreasing speed enormously but don't know. Is there anything can be better to preserve original quality or speed up encoding much at very minimal quality loss? I personally like aq mode 2 more in x264. I also noticed no early skip that seems to me also time consuming. Maybe there is something that is contrary to other setting x265 4:2:0

wpp / ctu=64 / min-cu-size=8 / max-tu-size=32 / tu-intra-depth=2 / tu-inter-depth=2 / me=3 / subme=4 / merange=57 / rect / amp / max-merge=3 /
temporal-mvp / no-early-skip / rskip / rdpenalty=0 / no-tskip / no-tskip-fast / strong-intra-smoothing / no-lossless / no-cu-lossless / no-constrained-intra /
no-fast-intra / open-gop / no-temporal-layers / interlace=0 / keyint=480 / min-keyint=24 / scenecut=40 / rc-lookahead=40 / lookahead-slices=4 / bframes=6 /
bframe-bias=0 / b-adapt=2 / ref=4 / limit-refs=3 / limit-modes / weightp / weightb / aq-mode=1 / qg-size=32 / aq-strength=1.00 / cbqpoffs=0 / crqpoffs=0 /
rd=4 / psy-rd=2.00 / rdoq-level=2 / psy-rdoq=1.00 / log2-max-poc-lsb=8 / no-rd-refine / signhide / deblock=0:0 / sao / no-sao-non-deblock / b-pyramid /
cutree / no-intra-refresh / rc=crf / crf=20.0 / qcomp=0.60 / qpmin=0 / qpmax=69 / qpstep=4 / ipratio=1.40 / pbratio=1.30
He is now using CRF 18. But I think this setting above, where used related to topic.

Any suggestion is welcomed.
Bernix