Poor H264 encoding quality in DaVinci

Get answers to your questions about color grading, editing and finishing with DaVinci Resolve.
  • Author
  • Message
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostThu Apr 06, 2017 6:49 pm

Al Spaeth wrote:Next - and a bit off topic - I have seen info claiming that h.265 HEVC produces better quality 1080p than h.264. I know it's half the file size as less than half the bitrate for same quality - but better quality??


Sorry, I missed this prior post. I don't think it is off topic really, but I will be brief on this one:

http://goughlui.com/2016/08/27/video-compression-testing-x264-vs-x265-crf-in-handbrake-0-10-5/

There you go - someone even more anal about metric testing....and long winded ;)

Edit:And some further bed-time reading:

http://www.compression.ru/video/codec_comparison/hevc_2015/
Last edited by Bryan Worsley on Thu Apr 06, 2017 11:32 pm, edited 1 time in total.
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostThu Apr 06, 2017 6:58 pm

Andrew Kolakowski wrote:
Bryan Worsley wrote:It's modern codec- quite different than others and this should work well in Resolve as it's GPU based. You read file, push data to GPU for decoding and Resolve processing.
I'm just worried it will die if big names won't support it (and they won't as there is no money there for them).
Cinegy is quite well known company in broadcast world, but this maybe not enough.


Interesting. Yeah Cinegy.....I seem to recall reading about this project several years back, under a different name maybe ?
Offline

Andrew Kolakowski

  • Posts: 9209
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Poor H264 encoding quality in DaVinci

PostThu Apr 06, 2017 7:05 pm

Last time it was called DANIEL :)
They should be talking about it at NAB.
Offline

Al Spaeth

  • Posts: 329
  • Joined: Thu Sep 17, 2015 9:48 pm
  • Location: South Africa

Re: Poor H264 encoding quality in DaVinci

PostThu Apr 06, 2017 7:28 pm

Cary Knoop wrote:
Al Spaeth wrote:Also saw a comparison of h.265 encoders HEVC vs VP9 and VP9 looks very good.
Why aren't NLEs using royalty free VP8 and VP9 for h.264 and h.265 more??
Al

I think VP8 is really not a player in the field.

VP9 is a great delivery codec, and, as you say, is royalty free. But VP9 is very processing heavy and completely unsuitable as an editing codec.


Thanks Cary - I meant as a delivery codec vs HEVCs heavy royalty costs.
Al
Resolve 15.3 free Win 10 64bit
Offline

Al Spaeth

  • Posts: 329
  • Joined: Thu Sep 17, 2015 9:48 pm
  • Location: South Africa

Re: Poor H264 encoding quality in DaVinci

PostThu Apr 06, 2017 7:35 pm

Andrew Kolakowski wrote:Last time it was called DANIEL :)
They should be talking about it at NAB.


Watch the video "Cinegy at NAB 2016"
Resolve 15.3 free Win 10 64bit
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostThu Apr 06, 2017 7:38 pm

Andrew Kolakowski wrote:Last time it was called DANIEL :)


*Looks for narrowed eyes emoji.....closest :| *

It must have been something different then.

Edit: It was this....Perseus from V-Nova:

http://www.digitaltveurope.net/346362/v-nova-promise-video-at-audio-bitrates-with-revolutionary-compression-technology/

http://www.v-nova.com/perseus/
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostFri Apr 07, 2017 6:14 pm

Bryan Worsley wrote:My inclination at this point - go with native 4K input and project, generate optimized media DNxHR-HQX at half-res (or less), and render out to DNxHR-HQX at original (project) res as your exchange intermediate for encoding to x264 with Handbrake or whatever. That's assuming of course that you do want to preserve 4K export.


But unfortunately, having just tested it, Handbrake won't open the DNxHR-HQX.mov files, at least the linux version I have installed (on Kubuntu). If you find the same, the options for transcoding DNxHR to x264, as I see it, would be:

FFMPEG - command line or front end. Best bet in my opinion.
MeGUI/AVISynth (with ffms2mod decode/file indexer). A bit tedious.
VirtualDub FilterMod with vfw version of x264 encoder.
Offline

Andrew Kolakowski

  • Posts: 9209
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Poor H264 encoding quality in DaVinci

PostFri Apr 07, 2017 6:40 pm

Make sure you have latest Handbrake- older version (<1.0) won't read DNxHR.
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostFri Apr 07, 2017 7:50 pm

I'm running v1.03 from:
https://launchpad.net/~stebbins/+archive/ubuntu/handbrake-releases
..and it won't open the DNxHR.mov files. But if the Windows version does, that's great. I only rarely use Handbrake, and then only in linux.

Edit: Checked out the Windows v1.03 version also and it won't open the Resolve DNxHR-HQX or HQ files:

Image
Offline

Andrew Kolakowski

  • Posts: 9209
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Poor H264 encoding quality in DaVinci

PostFri Apr 07, 2017 11:08 pm

Hmmm.. I thought it does work, but tried on Mac and it doesn't. Strange, as it should come with libav12 update.
You can use GrassValley HQX (also intermediate codec)- this should work.
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostSat Apr 08, 2017 12:44 am

Well it's not an issue for me, as I'm content using ffmpeg or AVISynth/MeGUI for x264 encoding. It's Al who would be stymied if he wanted to use a DNxHR 4K Resolve workflow with Handbrake for x264. I'd have to look at the through-put quality metrics again to satisfy myself that Grass Valley HQX could be a satisfactory alternative.

Incidentally, I find also that Handbrake 1.03 (Windows and Ubuntu versions) won't open Cineform (avi or mov) either, yet an earlier version (0.10.5) I had on Kubuntu did. It's how I did those Resolve QT H264 vs x264 tests with Cineform as the intermediate:

Bryan Worsley wrote:Here's an amended table including the Handbrake results:

https://www.datafilehost.com/d/9ecc654a


Edit: Considering alternatives with extensive configuration options.

MediaCoder (Basic)
http://www.mediacoderhq.com/

Also Hybrid:
http://www.selur.de/

Both import the DNxHR and Cineform files just fine.
Offline

Andrew Kolakowski

  • Posts: 9209
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Poor H264 encoding quality in DaVinci

PostSat Apr 08, 2017 11:08 am

Bryan Worsley wrote:Well it's not an issue for me, as I'm content using ffmpeg or AVISynth/MeGUI for x264 encoding. It's Al who would be stymied if he wanted to use a DNxHR 4K Resolve workflow with Handbrake for x264. I'd have to look at the through-put quality metrics again to satisfy myself that Grass Valley HQX could be a satisfactory alternative.

Incidentally, I find also that Handbrake 1.03 (Windows and Ubuntu versions) won't open Cineform (avi or mov) either, yet an earlier version (0.10.5) I had on Kubuntu did. It's how I did those Resolve QT H264 vs x264 tests with Cineform as the intermediate:

Bryan Worsley wrote:Here's an amended table including the Handbrake results:

https://www.datafilehost.com/d/9ecc654a


Edit: Considering alternatives with extensive configuration options.

MediaCoder (Basic)
http://www.mediacoderhq.com/

Also Hybrid:
http://www.selur.de/

Both import the DNxHR and Cineform files just fine.


HQX is as good as other codecs- just doesn't do 444, but if you going later to make h264 this doesn't mean much.

Check this:

http://s32.postimg.org/u54dzr079/test.png

It's properly done with special DCI testing clip (which was never compressed).
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostSat Apr 08, 2017 4:20 pm

Andrew Kolakowski wrote:
Bryan Worsley wrote:
Check this:

http://s32.postimg.org/u54dzr079/test.png

It's properly done with special DCI testing clip (which was never compressed).


Yes, a very interesting and informative analysis. Thanks for that Andrew. And a nice illustration of how Cineform does indeed cope with complex 'hard-to-compress' content in maintaining 'constant quality' (as judged by PSNR), albeit at the expense of variable bitrate.

If you don't mind me asking - did you do this analysis yourself, and if so, using what test system ?

Edit: Ah OK, I saw your earlier post:

https://forum.blackmagicdesign.com/viewtopic.php?f=3&t=48192
Offline

Al Spaeth

  • Posts: 329
  • Joined: Thu Sep 17, 2015 9:48 pm
  • Location: South Africa

Re: Poor H264 encoding quality in DaVinci

PostSat Apr 15, 2017 9:25 am

Bryan Worsley wrote:I'm quite new to Resolve myself, but it does pay to have a firm grasp of these formats and how they behave in the particular system you are working with.

On that note:

I am looking into this as and when I have time, largely because I too need to establish a manageable native 4K workflow. I'll post my findings and conclusions when I'm done, but there are some aspects of the behavior of Cineform and DNxHR that have come to light that I need to work through.

My inclination at this point - go with native 4K input and project, generate optimized media DNxHR-HQX at half-res (or less), and render out to DNxHR-HQX at original (project) res as your exchange intermediate for encoding to x264 with Handbrake or whatever. That's assuming of course that you do want to preserve 4K export. Don't even think about down-scaling as an HD project per se and rendering out 4K.


Thanks again Bryan - Please let us know what 4k workflow you find best for h.264 quality output.
Al
Resolve 15.3 free Win 10 64bit
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostSat Apr 15, 2017 6:47 pm

Al Spaeth wrote:
Bryan Worsley wrote:My inclination at this point - go with native 4K input and project, generate optimized media DNxHR-HQX at half-res (or less), and render out to DNxHR-HQX at original (project) res as your exchange intermediate for encoding to x264 with Handbrake or whatever. That's assuming of course that you do want to preserve 4K export. Don't even think about down-scaling as an HD project per se and rendering out 4K.


Thanks again Bryan - Please let us know what 4k workflow you find best for h.264 quality output.
Al


I can't really add anything more at this point Al; it's still very much 'work in progress'.
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostSun Apr 16, 2017 7:45 pm

Hey Andrew,
Re:
Bryan Worsley wrote:But unfortunately, having just tested it, Handbrake won't open the DNxHR-HQX.mov files, at least the linux version I have installed (on Kubuntu). If you find the same, the options for transcoding DNxHR to x264, as I see it, would be:

FFMPEG - command line or front end. Best bet in my opinion.
MeGUI/AVISynth (with ffms2mod decode/file indexer). A bit tedious.
VirtualDub FilterMod with vfw version of x264 encoder.


Just realized that VirtualDub FilterMod has it's own integrated x264 (8 and 10-bit) compressors i.e. doesn't require installation of the infrequently updated VFWx264 encoder:

https://forum.doom9.org/showthread.php?p=1803943#post1803943

Very cool.
(Edit: Appears to be an issue with the x264 compressors (not engaging) in the 32bit version of the latest VDubFilterMod release (#39148), but the 64bit version works fine)

Al, I think this could be a better option for you than Handbrake.
Last edited by Bryan Worsley on Mon Apr 17, 2017 3:29 am, edited 2 times in total.
Offline

Andrew Kolakowski

  • Posts: 9209
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Poor H264 encoding quality in DaVinci

PostSun Apr 16, 2017 8:13 pm

I don't need x264 GUIs, I use ffmpeg. Can even compile my own if need something custom :)
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostSun Apr 16, 2017 11:12 pm

Me too, unless I want to do some post-processing with AVISynth, in which case I tend to use MeGUI. I know, ffmpeg accepts AVS scripts also - it's just what I've got used to.

I was rather thinking of Al being limited to Grass Valley HQX as the export intermediate if he intends to use Handbrake for x264 encoding.
Offline

SteveMullen

  • Posts: 130
  • Joined: Thu Sep 15, 2016 3:08 am

Re: Poor H264 encoding quality in DaVinci

PostMon Apr 17, 2017 1:31 am

"e.g. do I really want to upload mp4's to YT when it gets re-encoded anyway?).

However, a reasonable question that I don't see asked very often is, if you are planning to use a third-party encoder, what format should you export? I have settled on image sequences, but I am curious what others say."

I've checked UT with PR422 (LT, std, HQ) plus AIC on the Mac.

On the PC; DNxHD (for HD), DNXHR (for 4K), Uncompressed 422.

Few reasons to export anything in H264.
Offline

Al Spaeth

  • Posts: 329
  • Joined: Thu Sep 17, 2015 9:48 pm
  • Location: South Africa

Re: Poor H264 encoding quality in DaVinci

PostMon Apr 17, 2017 7:47 am

Bryan Worsley wrote:Me too, unless I want to do some post-processing with AVISynth, in which case I tend to use MeGUI. I know, ffmpeg accepts AVS scripts also - it's just what I've got used to.

I was rather thinking of Al being limited to Grass Valley HQX as the export intermediate if he intends to use Handbrake for x264 encoding.


Hi Bryan - Thanks for the Vdub FilterMod tip. I saw x264 but assumed there was a reason you guys suggested HandBrake. I have the 64bit version so should be OK. As mentioned before, ffmpeg is way beyond my tech skills so I need a GUI like VDub - but even then, I would have never worked out how to set the correct Cineform 8bit/10bit pixel options recommended by Andrew above. (Any idea why when I choose YUY2 - it shows YUYV?)
I have a folder on my Desktop called Video Utilities and I copy the info/tips you guys have posted for reference - my version of "Resolve for Dummies". ;)

I thought h.264 AVC was a standard universal codec and any software encoder/decoder would produce the same results until I saw this thread. Then I saw this table on Wiki.

https://drive.google.com/open?id=0B9nToPFW14gZTkJqTmR0bHlkX0U
(How do I post a JPG image to the Forum????)
Quick Time seems one of the worst.

Slightly off topic - I was disappointed with h.264 mp4 render quality in another NLE I was using (Power Director) and noticed Media Info shows they are using a Sony Playstation codec. Can't find much info on it. Yet the
Format : MPEG-4
Format profile : Sony PSP
Codec ID : MSNV (MSNV/isom/mp42)
Format : AVC
Format/Info : Advanced Video Codec
Format profile : High@L5.1
Format settings, CABAC : Yes
Format settings, ReFrames : 2 frames
Format settings, GOP : M=3, N=13
Codec ID : avc1
Codec ID/Info : Advanced Video Coding

MSNV is shown as a QT file type - 'MSNV' : Sonys private brand; Used for example to encode MP4s for the PSP - Playstation was never known for it's video quality.
Any idea why they would choose to use this codec - and could it affect quality?

GoPro Studio mp4 MediaInfo
Format : MPEG-4
Format profile : Base Media
Codec ID : isom (isom/iso2/avc1/mp41)

Is there any way in MediaInfo to see what software/hardware encoder was used?
Resolve 15.3 free Win 10 64bit
Offline

Andrew Kolakowski

  • Posts: 9209
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Poor H264 encoding quality in DaVinci

PostMon Apr 17, 2017 12:12 pm

No- there is no reliable way to tell what software was used. Sometimes this info is stored sometimes is not and you can't tell at all.

All what you need are presets for ffmpeg+ simple GUI to tun it (or even just .bat file).
Vdub won't change much as you still need to set x264 parameters as in any other tool.
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostMon Apr 17, 2017 9:53 pm

Al Spaeth wrote: Thanks for the Vdub FilterMod tip. I saw x264 but assumed there was a reason you guys suggested HandBrake.


I for one didn't suggest Handbrake. You posed the question:

Al Spaeth wrote:What format should I choose for export from my timeline and then convert to h.264 mp4 using Handbrake's x264 encoder for best mp4 quality?


So it was reasonable to assume that Handbrake is what you are familiar with and want to use. And if you still do, that rules out Cineform and DNxHR as compatible UHD/4K export intermediates, at this time at least. But you still have Grass Valley HQX, which, as Andrew indicated, is a perfectly acceptable option; yes, I've done some tests.

Otherwise for Cineform or DNxHR you'll need to consider and get to grips with these other options we've discussed:

Al Spaeth wrote:As mentioned before, ffmpeg is way beyond my tech skills..

How do you know unless you try it. Installation is straightfoward. And once you've set the command lines for the different source formats you use and x264 formats you wish to encode to (which is not that different from configuring a GUI) you can save them as templates (presets) in a text file. Then its really just a matter of plugging in your input and destination (output) file path and names, and copy-pasting the line into a command prompt terminal. Not at all difficult once you get into it.

But if you are more comfortable using a GUI, fine, go for it. And if there are already presets for the particular x264 target formats you need, all the better.

Al Spaeth wrote:....so I need a GUI like VDub - but even then, I would have never worked out how to set the correct Cineform 8bit/10bit pixel options recommended by Andrew above. (Any idea why when I choose YUY2 - it shows YUYV?)


I think Andrew was talking there about transcoding your source footage to Cineform. YUYV is just another expression for YUY2. Granted though, using VDub FilterMod to convert the different Resolve 10-bit 422 export options to x264 does involve some consideration of the available decode formats and conversion paths ... I'm looking at that myself just now. And again, yes, you would need to configure the x264 encoder - there's no getting away from that, but once set-up you can save and re-load the processing settings, just like a preset.
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostTue Apr 18, 2017 4:18 pm

Bryan Worsley wrote:.... But you still have Grass Valley HQX, which, as Andrew indicated, is a perfectly acceptable option; yes, I've done some tests.

My last contribution on the subject:

https://www.datafilehost.com/d/7c980d99

The results speak for themselves. All of those x264 encodes are high quality. Note that for each export format I produced two sets of renders, one with Optimized Media (DNxHR-HQX at half res) turned on and the other turned off, and in each case the renders were bit-identical and produced identical SSIM scores - in other words applying Optimized Media does not of itself affect the (baseline) 'pass-through' quality efficiency when no transforms are applied.
Offline

Martin Schitter

  • Posts: 899
  • Joined: Tue Apr 28, 2015 10:41 pm

Re: Poor H264 encoding quality in DaVinci

PostTue Apr 18, 2017 5:09 pm

Bryan Worsley wrote:My last contribution on the subject...


indeed interesting again! :)

i think, your tests demonstrate one well known issue concerning cineform and other wavelet based compression formats. if they are used in cascaded compression scenarios, they work significant worse then DCT compression based sources, even if the quality and available bandwidth of the wavelet intermediates is higher. that's no surprise. it's a well know phenomenon. it's much better to stay in the same compression family, if you want to archive the best results over generations.
Offline

Andrew Kolakowski

  • Posts: 9209
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Poor H264 encoding quality in DaVinci

PostTue Apr 18, 2017 5:59 pm

Yes, that's the case and for that reason you should never do proper codec valuation starting with format which was already compressed as results maybe very misleading. Another thing to do is shift source file 2 pixels up/left which will break DCT blocks alignment.
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostTue Apr 18, 2017 7:39 pm

You will appreciate though that this was purely an 'in-use' test, of sorts, that I conducted primarily for my own reference in arriving at a manageable 4K workflow.

What it doesn't examine of course is the degree to which applying color transforms (i.e. the actual business of grading) to 8-bit 4:2:0 inputs via a 10-bit 422 'virtual intermediate' (the optimized media) benefits precision in the final outcome, and especially if the 10-bit 422 export format is different e.g. using DNxHR-HQX as the 'optimized medium' but exporting Cineform or Grass Valley HQX. How important is that the optimized media and render format are the same? And how does that compare to using a 10-bit 422 transcode of the 8-bit source for input? Rather more difficult to assess with simple metric tests like these.
Offline

Andrew Kolakowski

  • Posts: 9209
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Poor H264 encoding quality in DaVinci

PostTue Apr 18, 2017 8:09 pm

Yes, your testing is bit different.

In case of 8bit source there should be no difference if your optimised media is 10bit or 8bit (all Resolve formats are 10bit). It makes no difference as Resolve doesn't attempt to use any special conversion for 8<->10bit.

If you can set optimised media to format which you most likely going to export, so you can save time.
Offline

Al Spaeth

  • Posts: 329
  • Joined: Thu Sep 17, 2015 9:48 pm
  • Location: South Africa

Re: Poor H264 encoding quality in DaVinci

PostTue Apr 18, 2017 8:37 pm

Bryan Worsley wrote:My last contribution on the subject:

https://www.datafilehost.com/d/7c980d99

The results speak for themselves. All of those x264 encodes are high quality. Note that for each export format I produced two sets of renders, one with Optimized Media (DNxHR-HQX at half res) turned on and the other turned off, and in each case the renders were bit-identical and produced identical SSIM scores - in other words applying Optimized Media does not of itself affect the (baseline) 'pass-through' quality efficiency when no transforms are applied.


Thanks again Bryan -

Still having problems with my workflow

I tried MediaCoder x64 free but it seems to have a bitrate limitation of 16Mbps for x264 mp4 - if I understand it correctly.

Also found same Handbrake error "no valid source or titles found" when trying to open a Cineform AVi.

VirtualDub FilterMod x64 gives me an error "Internal error (FFMPEG): Invalid data found when processing input" (frame zero) on all Cineform AVI files I try to encode using x264 8bit or x264vfw (file type mp4) including AVIs from Resolve, GoPro Studio, and even those created by VirtualDub FilterMod. It imports and plays AVIs fine but bombs on encoding.
The same AVI clips work fine in MediaCoder barring the 16Mbps x264 limit - also can't see how to get MediaCoder to encode Cineform AVi (it does have DNxHD).

Please bear with me as I want to make sure I understand your test results.

All export codecs were in the 82-84% range and DNxHR HQX scored as high as Cineform Best=Filmscan 2 - what's your conclusion for best workflow?

Did DNxHR-HQX at half resolution give you smooth timeline performance vs your source mp4?

Why do you think your render results with Optimized Media on (half res) and off (rendered from your source clip) were identical?

If you still have rendered clips it would be interesting to compare file sizes.

Were your x264 mp4 clips similar bitrates to your source AVI mp4?

Would it be possible to add Resolve QT h.264 MOV, re-wrap to mp4 and compare SSIM% to show how bad Resolve h.264 is (our original topic).

Al
Resolve 15.3 free Win 10 64bit
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostTue Apr 18, 2017 11:59 pm

Al Spaeth wrote:I tried MediaCoder x64 free but it seems to have a bitrate limitation of 16Mbps for x264 mp4 - if I understand it correctly.

Didn't know that. I've never used it other than to test if it accepts Cineform and DNxHD. Maybe you need the paid Premium version to get higher bitrates.

Al Spaeth wrote:Also found same Handbrake error "no valid source or titles found" when trying to open a Cineform AVi.

We've already established that the current version of Handbrake does not accept Cineform or DNxHR input:

Bryan Worsley wrote:....So it was reasonable to assume that Handbrake is what you are familiar with and want to use. And if you still do, that rules out Cineform and DNxHR as compatible UHD/4K export intermediates, at this time at least. But you still have Grass Valley HQX, which, as Andrew indicated, is a perfectly acceptable option; yes, I've done some tests.

Otherwise for Cineform or DNxHR you'll need to consider and get to grips with these other options we've discussed


Next:

Al Spaeth wrote:VirtualDub FilterMod x64 gives me an error "Internal error (FFMPEG): Invalid data found when processing input" (frame zero) on all Cineform AVI files I try to encode using x264 8bit or x264vfw (file type mp4) including AVIs from Resolve, GoPro Studio, and even those created by VirtualDub FilterMod. It imports and plays AVIs fine but bombs on encoding.


I don't know why. Works fine for me. Have you tested Resolve DNxHR and Grass Valley HQX mov renders also ?

Al Spaeth wrote:...also can't see how to get MediaCoder to encode Cineform AVi (it does have DNxHD).

Probably because it can't. There's no mencoder/ffmpeg Cineform encoder, only decoder.

Al Spaeth wrote:All export codecs were in the 82-84% range and DNxHR HQX scored as high as Cineform Best=Filmscan 2 - what's your conclusion for best workflow?


You keep pressing me on this but I won't be drawn. The whole point of posting those results was to demonstrate just that - that all of those export formats are very high quality and from them you can produce very high quality x264 encodes. Have you tried Grass Valley HQX with Handbrake ? Like DNxHR, you won't be able to play the GV HQX renders with MPC-HC or VLC player, but MPV will play them very nicely.

Al Spaeth wrote:Did DNxHR-HQX at half resolution give you smooth timeline performance vs your source mp4?

Yes, it makes a difference. And what do you find ?

Al Spaeth wrote:Why do you think your render results with Optimized Media on (half res) and off (rendered from your source clip) were identical?


Because:

Bryan Worsley wrote: in each case the renders were bit-identical and produced identical SSIM scores.


Bit-identical - they had exactly the same file size down to the very last byte.

Al Spaeth wrote:If you still have rendered clips it would be interesting to compare file sizes.


Which you could also test yourself. No I don't still have the rendered clips. But the clip was 15.25 secs long. So if you multiply the reported bitrate (video stream) by that it will give you the approx file size minus the audio. But what more does that tell you any than comparing bitrates?

Al Spaeth wrote:Were your x264 mp4 clips similar bitrates to your source AVI mp4?


Bitrate of the source UHD/29.97 AVC mp4 clip was 91.0 Mbps (video stream) with 1536 kbps PCM audio.

Al Spaeth wrote:Would it be possible to add Resolve QT h.264 MOV, re-wrap to mp4 and compare SSIM% to show how bad Resolve h.264 is (our original topic).


Why ? We've already been through that.

Edit: But for that particular source, not bad actually at a comparable bitrate:
High - 144Mbps, SSIM 86.33
Medium - 83.9Mbps, SSIM 81.37
Low - 54Mbps, SSIM 77.16

OK, I'm done.
Last edited by Bryan Worsley on Wed Apr 19, 2017 4:34 am, edited 6 times in total.
Offline
User avatar

Frank Glencairn

  • Posts: 1801
  • Joined: Wed Aug 22, 2012 7:07 am
  • Location: Germany

Re: Poor H264 encoding quality in DaVinci

PostWed Apr 19, 2017 1:28 am

I usually stay uncompressed during post, including my master, from there I encode to the delivery format.

Worked just fine for me since years - fast and hassle free - no codec MumboJumbo
http://frankglencairn.wordpress.com/

I told you so :-)
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostWed Apr 19, 2017 2:25 am

Frank Glencairn wrote:I usually stay uncompressed during post, including my master, from there I encode to the delivery format.


There you go Al, there's your workflow ;)
Offline

Al Spaeth

  • Posts: 329
  • Joined: Thu Sep 17, 2015 9:48 pm
  • Location: South Africa

Re: Poor H264 encoding quality in DaVinci

PostWed Apr 19, 2017 8:17 am

Bryan Worsley wrote:
Frank Glencairn wrote:I usually stay uncompressed during post, including my master, from there I encode to the delivery format.


There you go Al, there's your workflow ;)


Sorry Bryan - I can see my lack of knowledge is annoying you - apologies - and I do appreciate your efforts. It is one of the dangers of allowing newbies using free Resolve to participate in a forum of highly skilled professional editors accustomed to a studio workflow.

Scripting (ffmpeg, etc) is just not an option for me and file sizes of 4k intermediates/uncompressed is a consideration for me. Master to uncompressed on a half hour project would create a significant storage/backup challenge and your results showed that there is no significant quality advantage over the intermediates you tested.

4k timeline performance is critical which means I need an intermediate for edit.

It appears my best 4k h.264 mp4 work-around using the Resolve options for now may be:
Source 4k h.264 8bit
> optimize to DNx variant/resolution for best timeline performance on my PC (I'll test with DNxHR HQX which, as Andrew suggested, should save render time)
> Render to DNxHR HQX
> Encode to h.264 mp4 using x264 GUI

The only difference to my current workflow is that I was rendering out to Resolve QT h.264 MOV and using VLC to re-wrap to mp4. I would still like to see this quality compared to Bryan's SSIM% results using x264 at a similar bitrate to his source.

My wishlist, as already suggested in the thread, is:
1) Resolve should add more delivery options including h.264 mp4 with an improved encoder.
2) Add better Cineform implementation as suggested by Andrew and proven in his test where he demonstrated smooth 4 layer 4k editing on a two core i5 cpu :shock: . I would think even the Pros would find it useful to edit and color correct 4k video with Resolve on a mobile i5 laptop :)

At the risk of further annoyance, I'll bow out now - thanks again to all of you.

Al
Resolve 15.3 free Win 10 64bit
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostWed Apr 19, 2017 11:19 am

Al Spaeth wrote:The only difference to my current workflow is that I was rendering out to Resolve QT h.264 MOV and using VLC to re-wrap to mp4. I would still like to see this quality compared to Bryan's SSIM% results using x264 at a similar bitrate to his source.


And I dutifully took the extra time to comply:

Bryan Worsley wrote:
Al Spaeth wrote:Were your x264 mp4 clips similar bitrates to your source AVI mp4?


Bitrate of the source UHD/29.97 AVC mp4 clip was 91.0 Mbps (video stream) with 1536 kbps PCM audio.

Al Spaeth wrote:Would it be possible to add Resolve QT h.264 MOV, re-wrap to mp4 and compare SSIM% to show how bad Resolve h.264 is (our original topic).


Why ? We've already been through that.

Edit: But for that particular source, not bad actually at a comparable bitrate:
High - 144Mbps, SSIM 86.33
Medium - 83.9Mbps, SSIM 81.37
Low - 54Mbps, SSIM 77.16


And therein lies the source of my frustration, not your knowledge level at all.
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostThu Apr 20, 2017 4:23 am

Bryan Worsley wrote:What it doesn't examine of course is the degree to which applying color transforms (i.e. the actual business of grading) to 8-bit 4:2:0 inputs via a 10-bit 422 'virtual intermediate' (the optimized media) benefits precision in the final outcome, and especially if the 10-bit 422 export format is different e.g. using DNxHR-HQX as the 'optimized medium' but exporting Cineform or Grass Valley HQX. How important is that the optimized media and render format are the same? And how does that compare to using a 10-bit 422 transcode of the 8-bit source for input? Rather more difficult to assess with simple metric tests like these.


Andrew Kolakowski wrote:
In case of 8bit source there should be no difference if your optimised media is 10bit or 8bit (all Resolve formats are 10bit). It makes no difference as Resolve doesn't attempt to use any special conversion for 8<->10bit.

If you can set optimised media to format which you most likely going to export, so you can save time.


Yes thanks Andrew. I've been running some tests with an uncompressed 8-bit 420 greyscale gradient clip and I see now that in order to reap the benefit of using 10-bit 422 optimised media in terms of the higher (internal) precision that brings to applied transforms, you must to render out the optimised media per se i.e. select 'Use Optimized Media' option in the Deliver>Render Settings. The precision is not transferable at render. I think that's right ? That's what the tests seemed to show any way.

Put another way, yes optimised media can be used in place of a 10-bit 422 transcode of an 8-bit 420 source, but if you don't render out the optimised media itself the outcome will be just the same as if you used the original 8-bit source for input, regardless of how the scopes look (smoother histogram profiles for one) when 'Use Optimized Media if Available' is selected in the Playback options - in which case, the only potential benefit is in timeline performance.

I think that is what you meant there by:

Andrew Kolakowski wrote:If you can set optimised media to format which you most likely going to export, so you can save time.


Export in that case meaning the optimised media itself, not setting the render format to match optimised media format ?

If I'm wrong about that, I'd be more than happy to be put straight, as this profoundly impacts my envisaged workflow for UHD/4K 8-bit 420 sources.

Edit: After further testing, I've changed my view on that. With 8-bit 420 inputs, exporting to a 10-bit 422 format (say DNxHR-HQX) does ANYWAY make for higher precision in applied color transforms ('grades'), irrespective of whether optimized media are generated (in the same format and resolution) and applied on the timeline or not. Furthermore, exporting the optimised media itself does not improve on that. So really the only benefit in generating 10-bit 422 optimized media for 8-bit 420 sources is for timeline performance....as well as the gratification of seeing smoother scope profiles on the timeline as transforms are applied. That's good.
Last edited by Bryan Worsley on Sun Apr 23, 2017 10:03 pm, edited 1 time in total.
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostSun Apr 23, 2017 8:00 pm

A bit off-topic I know but....going on to test native UHD AVC (8-bit 4:2:0) footage with applied grades, what I am seeing is clipping when highlight pull-down is applied to optimized media (DNxHR-HQX or Uncompressed 422 10-bit) at Full Data Levels. What's interesting is that the clipping only appears on the timeline; the rendered optimized media shows no clipping. Furthermore there is no clipping when a transcode of the native clip in the same 10-bit 422 format is used for input. So why only with optimized media on the timeline ?

Not such a big deal when the intent is to bring-down near blown highlight detail to 'broadcast safe' (video) levels, but sometimes you might want to pull a luma qualifier to preserve specular highlights and the clipping gets in the way of that. I learn that the clipping can be avoided using Uncompressed 16-bit float as the optimized media, but that seems like over-kill for 8-bit footage.
Offline

Andrew Kolakowski

  • Posts: 9209
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Poor H264 encoding quality in DaVinci

PostMon Apr 24, 2017 9:58 am

Bryan Worsley wrote:
Bryan Worsley wrote:
Edit: After further testing, I've changed my view on that. With 8-bit 420 inputs, exporting to a 10-bit 422 format (say DNxHR-HQX) does ANYWAY make for higher precision in applied color transforms ('grades'), irrespective of whether optimized media are generated (in the same format and resolution) and applied on the timeline or not. Furthermore, exporting the optimised media itself does not improve on that. So really the only benefit in generating 10-bit 422 optimized media for 8-bit 420 sources is for timeline performance....as well as the gratification of seeing smoother scope profiles on the timeline as transforms are applied. That's good.


Converting 8bit to 10bit is not going to make your footage "more grading resistant" or give any advantage in Resolve in terms of quality. This would only be true for some advanced 8bit->10bit conversion, but this is not what Resolve does.
Offline

Al Spaeth

  • Posts: 329
  • Joined: Thu Sep 17, 2015 9:48 pm
  • Location: South Africa

Re: Poor H264 encoding quality in DaVinci

PostMon Apr 24, 2017 10:11 am

Andrew Kolakowski wrote:
In case of 8bit source there should be no difference if your optimised media is 10bit or 8bit (all Resolve formats are 10bit). It makes no difference as Resolve doesn't attempt to use any special conversion for 8<->10bit.

If you can set optimised media to format which you most likely going to export, so you can save time.


Andrew Kolakowski wrote:
Converting 8bit to 10bit is not going to make your footage "more grading resistant" or give any advantage in Resolve in terms of quality. This would only be true for some advanced 8bit->10bit conversion, but this is not what Resolve does.


Thanks Andrew - If I understand you correctly - for 8 bit source use 8 bit optimised media and export same 8 bit format for best quality and faster render??
Resolve 15.3 free Win 10 64bit
Offline

Andrew Kolakowski

  • Posts: 9209
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Poor H264 encoding quality in DaVinci

PostMon Apr 24, 2017 11:02 am

You can use 10bit optimised media format (there is no 8bit one in Resolve) for 8bit source, but it won't make footage any "better" for grading.
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostMon Apr 24, 2017 12:52 pm

DNxHR-HQ and SQ are both 8-bit 422 aren't they ?

More to the point, perhaps the question that should be asked is if it is in any way detrimental to convert 8-bit 420 sources (UHD/4K or HD) to a 10-bit 422 format for grading in Resolve and/or export in a 10-bit 422 format for the purpose of transcoding to 8-bit 420 H264 targets externally ?

If it is preferable to "stay in 8-bit" throughout, that limits the '8-bit intermediate' format options available for import and export, even if optimized media are not used.

And if does do 'no harm' to grade and/or export as 10-bit 422 consideration still needs to be given to how it is converted back to 8-bit 420 by or for the external H264 encoder in question. One of the reasons why I've stayed with Cineform.avi (for the most part) as an intermediate for my HD-AVC material (not just with Resolve) and using AVISynth/MeGUI as the "x264 encoder system" is because Cineform could be relied on to perform the internal conversions to and from YUY2 properly and efficiently, and likewise AVISynth for the YUY2 <->YV12 conversion.

Edit:
Bryan Worsley wrote:A bit off-topic I know but....going on to test native UHD AVC (8-bit 4:2:0) footage with applied grades, what I am seeing is clipping when highlight pull-down is applied to optimized media (DNxHR-HQX or Uncompressed 422 10-bit) at Full Data Levels. What's interesting is that the clipping only appears on the timeline; the rendered optimized media shows no clipping. Furthermore there is no clipping when a transcode of the native clip in the same 10-bit 422 format is used for input. So why only with optimized media on the timeline ?


The clipping also occurs with DNxHR-HQ by the way so (assuming DNxHR-HQ is indeed 8-bit) it can't be a 10-bit thing specifically. I'm still interested to know why it only appears on the timeline with optimized media and not with transcodes....at least transcodes that preserve the full luma range of the source and don't suffer from clipping or compression in the process:

https://forum.blackmagicdesign.com/viewtopic.php?f=21&t=55098&start=100#p334584

Which in the context of DNxHR means transcoding externally with ffmpeg.
Offline

Andrew Kolakowski

  • Posts: 9209
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Poor H264 encoding quality in DaVinci

PostMon Apr 24, 2017 3:27 pm

Bryan Worsley wrote:DNxHR-HQ and SQ are both 8-bit 422 aren't they ?


Sorry, yes they are, but it doesn't really matter if you use 8bit intermediate or 10bit. Calculations are done at 32bti float, so data once decode gets always converter to this format and then back to some other when send to the codec for the export.
Resolve is not a transcoder, where you can avoid some not needed conversion steps (e.g. RGB<->YUV or 8bit<->10bit). Resolve has to normalise all incoming data to 32bit float format.
As I already mentioned for this reason Resolve is not the best choice when you need to convert YUV based formats to other YUV based ones as it will always force conversion to 32bit float 4:4:4.
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostMon Apr 24, 2017 11:56 pm

OK thanks Andrew. Time for some reflection on my existing HD and proposed 4K/UHD workflows I think.

BTW - I see Resolve 14 beta has brought 10-bit flavors of DNxHR-HQX and DNxHR-444 in addition to 12-bit, as well as the option for CBR or VBR encoding. Can I face running more tests - maybe not right now ;)
Offline

Al Spaeth

  • Posts: 329
  • Joined: Thu Sep 17, 2015 9:48 pm
  • Location: South Africa

Re: Poor H264 encoding quality in DaVinci

PostTue Apr 25, 2017 1:14 am

Bryan Worsley wrote:OK thanks Andrew. Time for some reflection on my existing HD and proposed 4K/UHD workflows I think.

BTW - I see Resolve 14 beta has brought 10-bit flavors of DNxHR-HQX and DNxHR-444 in addition to 12-bit, as well as the option for CBR or VBR encoding. Can I face running more tests - maybe not right now ;)


Played with 14 last night (till 3am).
14 has h.264 mp4 Deliver option (finally).
Would be interesting to compare quality of your h.264 mp4 8bit source > h.264 mp4 Deliver (Quality Auto Best) vs Deliver DNxHR-HQX > x264 mp4.
Much faster h.264 timeline performance too.

• Improved H.264 encode and decode performance
• Support for H.264 multi-pass encoding on Mac
• Support for decoding H.264 Intra 422 10bit - DaVinci Resolve Studio only
• Support for encoding to the MP4 format
• Support for decoding High 10 Profile H.264 files on DaVinci Resolve Studio
• Support for encoding VP9 profiles with higher bit-depth
Resolve 15.3 free Win 10 64bit
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostTue Apr 25, 2017 1:56 am

Al Spaeth wrote:14 has h.264 mp4 Deliver option (finally)
Would be interesting to compare quality of your h.264 mp4 8bit source > h.264 mp4 Deliver (Quality Auto Best) vs Deliver DNxHR-HQX > x264 mp4.
Much faster h.264 timeline performance too.


I'm aware of the mp4 render option, but it makes no difference if the container is mov or mp4 - the quality of the encoded h264 elementary stream (essence) is the same. And I already posted the results of that test with Resolve h264.mov

Yes I do see an improvement in native h264 timeline performance, but I've yet to see how that holds up under heavy grades.

Al Spaeth wrote:• Support for decoding H.264 Intra 422 10bit - DaVinci Resolve Studio only
....
• Support for decoding High 10 Profile H.264 files on DaVinci Resolve Studio


Disappointing and puzzling that it's only the Studio version that gets this.
Offline

Al Spaeth

  • Posts: 329
  • Joined: Thu Sep 17, 2015 9:48 pm
  • Location: South Africa

Re: Poor H264 encoding quality in DaVinci

PostTue Apr 25, 2017 8:15 am

Bryan Worsley wrote:
Al Spaeth wrote:14 has h.264 mp4 Deliver option (finally)
Would be interesting to compare quality of your h.264 mp4 8bit source > h.264 mp4 Deliver (Quality Auto Best) vs Deliver DNxHR-HQX > x264 mp4.
Much faster h.264 timeline performance too.


I'm aware of the mp4 render option, but it makes no difference if the container is mov or mp4 - the quality of the encoded h264 elementary stream (essence) is the same. And I already posted the results of that test with Resolve h264.mov

Yes I do see an improvement in native h264 timeline performance, but I've yet to see how that holds up under heavy grades.

Al Spaeth wrote:• Support for decoding H.264 Intra 422 10bit - DaVinci Resolve Studio only
....
• Support for decoding High 10 Profile H.264 files on DaVinci Resolve Studio


Disappointing and puzzling that it's only the Studio version that gets this.


Looks like the days of free Resolve are coming to an end which I expected - but for serious users, they have slashed the price of Studio. Maybe they will offer a 30 day trial so you can test the Studio h.264 features and new Studio only effects.
14 free still offers far more than any other free NLE out there (the what's new list is mind boggling) and the Studio price cut is significant. I think their marketing has been brilliant introducing a new NLE into an established Pro market :)

• Improved H.264 encode and decode performance
The h.264 encoder has changed. I thought it just might be worth checking MP4 quality against 12.5.5 MOV which is the basis for this thread. Odds are it's the same but it may be good to check before we continue this thread.
Resolve 15.3 free Win 10 64bit
Offline

Andrew Kolakowski

  • Posts: 9209
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Poor H264 encoding quality in DaVinci

PostTue Apr 25, 2017 9:46 am

On a Mac Resolve most likely uses Apple h264 encoder (based on the options naming and features).
I think BM doesn't want to pay h264 licensing fees, so they use what is provided by OSes to avoid additional costs.
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostTue Apr 25, 2017 11:23 am

Al Spaeth wrote:The h.264 encoder has changed. I thought it just might be worth checking MP4 quality against 12.5.5 MOV which is the basis for this thread. Odds are it's the same but it may be good to check before we continue this thread.


I've rolled back to 12.5.5. Experienced some crashes with the 14 beta - "DaVinci Resolve has stopped working etc" - and I'm not particularly inclined to delve into why. Still chapped about the omitted support for Pana GH5 10-bit files in the free version. And how are you supposed to evaluate whether it's worth paying for the Studio version if there's no trial download version of it ?

But feel free to check the mp4 vs mov H264 quality for yourself if it matters to you. It doesn't to me.
Offline

Al Spaeth

  • Posts: 329
  • Joined: Thu Sep 17, 2015 9:48 pm
  • Location: South Africa

Re: Poor H264 encoding quality in DaVinci

PostTue Apr 25, 2017 12:12 pm

Bryan - Just suggesting a quality test of ver 14 h.264 encoding vs your 12.5 results.
Thought that was the purpose of this thread :?:
Resolve 15.3 free Win 10 64bit
Offline
User avatar

Noel Sterrett

  • Posts: 521
  • Joined: Wed Aug 22, 2012 9:12 pm
  • Location: Atlanta

Re: Poor H264 encoding quality in DaVinci

PostTue Apr 25, 2017 1:11 pm

Resolve Studio 12.5 can export the Sony XAVC Intra VBR and CBG codecs with bitrates from 100 to the new 480. They are all H.264, and can be used by both Resolve versions.

I've been comparing the 480 to RAW on an F5, and it is so good that in most cases I will likely use it instead of RAW.

RAW needs less CPU for the decode, and as a result plays a bit smoother on the timeline than 480 in 12.5, but it is likely 14 will change that.

Cheers.
Admit One Pictures
Resolve Studio 18 | Linux Lint 21 | Nvidia 515 | Xeon | iCore | Ryzen
Offline

Bryan Worsley

  • Posts: 513
  • Joined: Fri Apr 15, 2016 11:26 am
  • Location: Montreal, Canada

Re: Poor H264 encoding quality in DaVinci

PostTue Apr 25, 2017 1:28 pm

Al Spaeth wrote:Bryan - Just suggesting a quality test of ver 14 h.264 encoding vs your 12.5 results.
Thought that was the purpose of this thread :?:


I know that's what you were suggesting and had I not experienced these crashes (which I recall occurred when I was manipulating some UHD AVC files) I would likely have gone on to test just that. Note though that it claims :

• Improved H.264 encode and decode performance

i.e. efficiency = speed. No mention of improved quality.

Plus, like I said before, I think you are hanging too much on these test results I posted. As I have been keen to point out these were 'in-use' tests (controlled as best) conducted primarily for my own purposes, which I thought might be of general interest. They were conducted with just one test clip in each case (as representative of the sources I am interested in) using just one quality measure (that I find particularly useful) and using an unrestricted x264 encode profile that I considered allowed valid comparison under those test conditions, which does not necessarily represent the target profiles that may be chosen in practice.

What's important is that you come to a conclusion as to which of the possible workflows that have been suggested and are feasible for you produce the best H264 quality, in your eyes, and for the particular delivery platforms you are targeting.

Also, I for one am not a 'Big Studio Guy'. I'm just someone like you, relatively new to Resolve and looking to get the best from it. I certainly don't want be held as some in-house 'quality tester'. I'm just "one b*gger telling another b*gger where they can find bread", so to speak.

Glad you are happy with Resolve 14 beta though:

https://forum.blackmagicdesign.com/viewtopic.php?f=21&t=58578#p336375

Love the "So far, it's now stable (I don't know why)" as if that's something to be suspicious of :lol:
Offline

Al Spaeth

  • Posts: 329
  • Joined: Thu Sep 17, 2015 9:48 pm
  • Location: South Africa

Re: Poor H264 encoding quality in DaVinci

PostTue Apr 25, 2017 2:37 pm

Bryan - Also had a few problems with 14. Crashed randomly Win 10 on first install.
Uninstalled 12.5.5 and 14 > rebooted > reinstalled 14 "run as Admin" and was much better. Got a Resolve error "GPU memory full" with two mp4 UHD layers no effects on the timeline (Nvidia 4GB GPU) > restarted 14 > preferences > GPU Auto

Simplest workflow.
I did a two layer UHD edit with color correction and PIP > Render from camera source UHD h.264 AVC High@L5.2 to UHD mp4 (233Mb) best quality and again to 1080p (49.7Mb) best.
Very happy with quality/detail of both but it's a subjective judgement call. UHD bitrate seems high as source UHD is only 22Mb/s so I'll change best quality to a lower bitrate (30Mb) and hopefully same quality and smaller file size.

Next:
Same project > Opt Media DNxHR HQX 1/4 > Render from source (Haven't tried Render from Opt Media)
First edit timeline OK and faster render times
Second edit timeline and scrubbing smooth, preview good, faster DNxHR Optimize than 12.5.

Stable now (I don't know why) but New Features list is mind boggling so I assume beta testing will take a while.

My h.264 workflow problem was 12.5 UHD timeline performance, slow transcoding, and no mp4 render out. 14 has greatly improved all.

So I'll stay with my simple 8 bit workflow and await your empirical results to see how much better x264 encoding is from DNx/Cineform vs Resolve 14 mp4. Thanks :)
Resolve 15.3 free Win 10 64bit
PreviousNext

Return to DaVinci Resolve

Who is online

Users browsing this forum: 4EvrYng, Bing [Bot], Google [Bot], iandunham, shikawkee, Simnut and 173 guests