Why do some people use Huffyuv,despite it's slower & bad compression ratio

[Truthler]

https://www.redsharknews.com/technology-computing/item/1183-why-can-t-there-just-be-on...lly-good-codec

Why do some people still cling on HuffYuv? It is at least 40% slower (with higher CPU usage) than Ut Video and it has 10-15% worse compression ratio.

[richardpl]

I have different experience with FFmpeg implementation of huffyuv, its is on par with utvideo maybe even sometimes faster.

[Truthler]

I have different experience with FFmpeg implementation of huffyuv, its is on par with utvideo maybe even sometimes faster.

Please test the Utvideo with median option and Huffyuv . Ut video is at least 40% faster and has 10-15% smaller file size.

[richardpl]

I did, and it is much slower than huffyuv implementation in ffmpeg, but size is bigger than utvideo as expected.

[lordsmurf]

Most of your posts at VH are extremely basic newbie questions, so I don't think you have the experience or qualifications to determine what is and isn't good quality for software, codecs, etc.

Utvideo is not better. It has quality issues, it has resource overhead issues (ie, dropped frames for capture, slow I/O for non-capture).

The ffmpeg version of Huffyuv is inferior to the original as well -- for capturing. It has loading issues in various apps that otherwise see original Huffyuv (MainConcept, for example). It's Huffyuv, but it also really isn't.

For an overall intermediary, for SD -- NOT CAPTURE! -- Lagarith on Windows is best, ProRes422 on Mac is best (HQ, and ProRes422 is lossy, yes, but not very; Utvideo is still worse).

For HD, FFv1 and ProRes again.

Your constant "Why don't you use new stuff??? WAH!!!!" posts are juvenile and clueless. If you actually did video, instead of just spouting off about things for which you know not, then you'd see and experience the issues were refer to.

[Truthler]

I did, and it is much slower than huffyuv implementation in ffmpeg, but size is bigger than utvideo as expected.

I don't know which softwares did you use. Did you use android mobil phone....

Try OBS or Virtualdub2. Install Utvideo codecs for VD2.

[Truthler]

Most of your posts at VH are extremely basic newbie questions, so I don't think you have the experience or qualifications to determine what is and isn't good quality for software, codecs, etc.

Utvideo is not better. It has quality issues, it has resource overhead issues (ie, dropped frames for capture, slow I/O for non-capture).

The ffmpeg version of Huffyuv is inferior to the original as well -- for capturing. It has loading issues in various apps that otherwise see original Huffyuv (MainConcept, for example). It's Huffyuv, but it also really isn't.

For an overall intermediary, for SD -- NOT CAPTURE! -- Lagarith on Windows is best, ProRes422 on Mac is best (HQ, and ProRes422 is lossy, yes, but not very; Utvideo is still worse).

For HD, FFv1 and ProRes again.

Your constant "Why don't you use new stuff??? WAH!!!!" posts are juvenile and clueless. If you actually did video, instead of just spouting off about things for which you know not, then you'd see and experience the issues were refer to.

It is time to use new versions, not the original 2008 version. 12years passed! Good Morning!
Use OBS or VD2 for the test. For transcoding, use Hybrid, it gives 140 FPS for 4K video files with i9-10850K CPU overclocked to 4.4GHZ.
Regarding 4K video transcoding with Huffyuv I got only 80-90FPS.
Prores is lossy, I don't record/capture edit anything in losslesss mode. I won't use lossy codec to store my camera's 4K videos.
FFV1 is only an archiving format, you can not play it back in normal speed in a video player. My CPU is very slow to play FFV1 in 4K

[richardpl]

Most of your posts at VH are extremely basic newbie questions, so I don't think you have the experience or qualifications to determine what is and isn't good quality for software, codecs, etc.

Utvideo is not better. It has quality issues, it has resource overhead issues (ie, dropped frames for capture, slow I/O for non-capture).

The ffmpeg version of Huffyuv is inferior to the original as well -- for capturing. It has loading issues in various apps that otherwise see original Huffyuv (MainConcept, for example). It's Huffyuv, but it also really isn't.

For an overall intermediary, for SD -- NOT CAPTURE! -- Lagarith on Windows is best, ProRes422 on Mac is best (HQ, and ProRes422 is lossy, yes, but not very; Utvideo is still worse).

For HD, FFv1 and ProRes again.

Your constant "Why don't you use new stuff??? WAH!!!!" posts are juvenile and clueless. If you actually did video, instead of just spouting off about things for which you know not, then you'd see and experience the issues were refer to.

You are very confused here. Codec software implementation have nothing to do with capture software implementation, one needs to spot that big difference. It is known fact that ffmpeg capture for windows is not optimized in any way and not much maintained.

Lagartith use floats for operation, and is really slow and supports only few formats. MagicYUV is much better. And also supports myriad more pixel formats like ffvhuff codec (ffmpeg version of huffyuv) in ffmpeg.
Also you lack almost any expertise to comment in such stuff in general.

[lordsmurf]

You are very confused here. Codec software implementation have nothing to do with capture software implementation,

I'm not confused whatsoever.
codecs =! capture software ... and I don't see how you thought I mushed those concepts together?

Lagartith use floats for operation, and is really slow and supports only few formats.

Formats?
When it comes to SD analog consumer formats (VHS, Hi8, etc), even a few pro analog formats, what do you think is missing?

MagicYUV is much better.

Not really. Tried it, have it, bought it, don't use it.
The ideal codec toolbox for SD video is Huffyuv (capture), Lagarith (intermediary), ProRes422 (if Mac work needed).

Also you lack almost any expertise to comment in such stuff in general.

Amusing. I've been capturing video for 20 years, and have taught millions of people how to also capture in that time. My methods are quality, not just using random hardware, codecs, software, and software settings.

[poisondeathray]

Utvideo is not better. It has quality issues, it has resource overhead issues (ie, dropped frames for capture, slow I/O for non-capture).

Are you referring the to VFW/ACM version ?

What "quality" issues ?

Yes, some people have reported dropped frames in some VFW/ACM capture scenarios . It does have higher % in that case. You might drop frames on an old computer . A 10 year old dual core should be sufficient for SD capture. YMMV - some people report no issues for HW capture using UT.

When it comes to SD analog consumer formats (VHS, Hi8, etc), even a few pro analog formats, what do you think is missing?

Likely he's referring to digibeta, 10bit422 SDI, broadcast, professional work

[Truthler]

Utvideo is not better. It has quality issues, it has resource overhead issues (ie, dropped frames for capture, slow I/O for non-capture).

Are you referring the to VFW/ACM version ?

What "quality" issues ?

Yes, some people have reported dropped frames in some VFW/ACM capture scenarios . It does have higher % in that case. You might drop frames on an old computer . A 10 year old dual core should be sufficient for SD capture. YMMV - some people report no issues for HW capture using UT.

When it comes to SD analog consumer formats (VHS, Hi8, etc), even a few pro analog formats, what do you think is missing?

Likely he's referring to digibeta, 10bit422 SDI, broadcast, professional work

I searched my old core2duo laptop in the cellar, and it can capture 1080p@24 with Utvideo.

[poisondeathray]

I searched my old core2duo laptop in the cellar, and it can capture 1080p@24 with Utvideo.

I don't have any issues with it either;

A handful of people have reported issues capturing with it over the years, for whatever reason - even though nobody else can replicate those issues on similar hardware. It's one of those mysteries that have never been solved. Believe me, people have tried looking into it.

Bottom line, use whatever works for your scenario .



There is a larger delta in performance for newer computer vs. older because of SIMD optimizations. eg SSE4.1/AVX/AVX2, etc.. classic huffyuv ACM does not use any of those for speedup, and older computers don't have them and cannot benefit either when using newer codecs like UT

People have to be clear about which version they are using for encoding/decoding when making comments or observations. What hardware, what scenario, what software are being used. There are many versions of huffyuv. There are significant differences between VFW/ACM version of UT and ffmpeg/libav version . For example most NLE's will be using a ACM version to decode, but not opensource ones like shotcut or blender

[Truthler]

I searched my old core2duo laptop in the cellar, and it can capture 1080p@24 with Utvideo.

I don't have any issues with it either;

A handful of people have reported issues capturing with it over the years, for whatever reason - even though nobody else can replicate those issues on similar hardware. It's one of those mysteries that have never been solved. Believe me, people have tried looking into it.

Bottom line, use whatever works for your scenario .



There is a larger delta in performance for newer computer vs. older because of SIMD optimizations. eg SSE4.1/AVX/AVX2, etc.. classic huffyuv ACM does not use any of those for speedup, and older computers don't have them and cannot benefit either when using newer codecs like UT

People have to be clear about which version they are using for encoding/decoding when making comments or observations. What hardware, what scenario, what software are being used. There are many versions of huffyuv. There are significant differences between VFW/ACM version of UT and ffmpeg/libav version . For example most NLE's will be using a ACM version to decode, but not opensource ones like shotcut or blender

Yes I agree. Ut video also has AVX/AVX2 support, so with new CPUs you get even faster encoding speed.
Read the AVX support: http://umezawa.dyndns.info/archive/utvideo/utvideo-5.1.2-readme.en.html
HuffYuv is slower due to the lack of AVX support. Don't forget, Utvideo was started as a HuffYuv upgrade... We could call it as Huffyuv 2.0

[ItaloFan]

https://www.redsharknews.com/technology-computing/item/1183-why-can-t-there-just-be-on...lly-good-codec

Why do some people still cling on HuffYuv? It is at least 40% slower (with higher CPU usage) than Ut Video and it has 10-15% worse compression ratio.

[I'm responding to the first post in the thread, since no one mentioned reading the linked article...]

OK, so I read the poorly edited opinion piece you linked to, and I did not get from it what you seem to have gotten from it.

The author's initial question in the headline ("Why can't there just be one really good codec?") laments the fact that there isn't yet one ideal intermediate/"mezzanine" lossless codec. He mentions that he got to thinking about this because someone from Adobe told him they get a lot of requests to make a codec to compete with ProRes and DNxHD. Adobe is not interested in doing that.

He then talks positively about both HuffyYUV/Lagarith (which he lumps together) and Ut Video. He describes entropy and Huffman coding as being challenges, and says he notices Ut Video does pretty well with that. However, he doesn't make any comparisons of Ut Video against HuffyYUV/Lagarith. It seems he is not really advocating it for any reason other than that it is newer and actively maintained. He even says he is disappointed in that it does not have 10-bit support.

In the final section of the article, he answers the question by saying basically that these codecs just leapfrog each other with support for the latest video standards, and that this type of codec may not even be the answer, given that uncompressed video handling may soon be within reach.

So, this article is far from being an endorsement of Ut Video over HuffyYUV and Lagarith for any technical reasons whatsoever. It is not really supporting your argument at all. Why do you even care what codec people use for their own files? You could ask the same rhetorical questions for audio: "why do people cling to FLAC when they could be using APE?" or "why cling to 48 kHz stereo for capture when the content is bandlimited mono?" You could maybe get some thoughtful answers if you were serious, but I agree with everyone else here; you seem to be just picking fights, not asking serious questions or really trying to understand what reasons people might have for not using your favorite codecs and software.

[poisondeathray]

He even says he is disappointed in that it does not have 10-bit support.

He should fact check his disappointment

The date of that article was Nov 24,2019.

Initial 10bit422 support was added 5 years before that article was published (for the VFW/ACM version ; the ffmpeg/libav version does not support 10bit even today), and 10bit RGB in 2016

UT Codec Version History
Version 14.0.0 	2014-04-15

New features

        Add a codec whose internal format is YUV422 10bit. (FourCC: UQY2) Very slow.


Version 16.0.0   2016-04-23

New features

        Add codecs whose internal formats are RGB 10bit and RGBA 10bit. (FourCC: UQRG, UQRA). Very slow.

That is an opinion article, but many benchmarks have been published before on other sites re: lossless codecs , decoding speeds, encoding speeds, pixel formats