-
Videoon 7/30/2021
Vegas Pro 18 build 527 not fully using GPU
@Richard-Green I believe there are two near separate things at play here, which can easily get conflated but shouldn’t be. #1 Todds point, which I also mentioned, your i5. Your machine will not render out the same clip in the same time as a machine with a faster Cpu, because although it’s a HW Nvenc encode the Cpu is involved in the render processing chain as well. #2 At this stage no one here I believe disputes that VP renders HW Nvenc slower than say Resolve. Personally, if I had to choose between getting the 3x faster render with Resolve or the 11% higher quality output with VP, i’d choose VP. But thats just me, where I put quality high on my list. In reality, if not constrained by file size output, I would simply increase the Resolves file size and data rates to match the VP quality, getting a still much faster render. But I don’t use Resolve, so for me I haven’t that choice, and I hope that VP utilises the Nvidia hardware as well as it does the Cpu going forward.
-
Videoon 7/29/2021
Vegas Pro 18 build 527 not fully using GPU
@Richard-Green Hi Richard "My original question is why isn't Vegas using the GPU as much as it should be" The only people that can answer that are the developers, you and I can only guess. It definitely looks from all of your testing that there's room for improvement in the Nvenc render speed capability. Best to wait and see if VP19 improves things, fingers crossed. I tweaked the results above of the three tested output files and added a S/S screenshot with a % column. The 100% value is simply a starting value so as to get a % difference between the other two, it doesn't imply that the 100% item is a 100% quality result, its a convenience.
-
Videoon 7/29/2021
Vegas Pro 18 build 527 not fully using GPU
@Richard-Green “No matter what I do I can't get it any faster without losing the quality” I was looking through the whole thread for your machines specs, now I see them in your last post🤪. It’s a good idea to put them in your profile, then easy for everyone to get at when doing this kind of testing. I suspect the the reason you’re not going to improve your render speed using nvenc is the i5. The CPU still obviously is part of the render chain. In other words if you had a faster Cpu, even using Nvenc I think you’re render speed would improve. I suspect that the Gpu spends some time waiting on your Cpu. The ideal is a balanced system, whatever that is🤣. Your Cpu usage is around 80%, my Cpu usage a lot less.
-
Videoon 7/29/2021
Premiere vs Vegas - 4K video decode/preview video cross track test
@BruceUSA I was wondering why I hadn't heard from you in quite some time. That's a pity really, you've been around here long before I started with Platinum 13. I agree. There is another way I look on it though, when the Fanboys descend, I just stay and annoy them more. Maybe it's a combination of fanboy and Stockholm Syndrome. Anyway, don't let them get you down, great to hear from you again.
-
Videoon 7/29/2021
Premiere vs Vegas - 4K video decode/preview video cross track test
Welcome sentiments. But I have seen moderators close down discussions about Resolve. It was a while back so I cannot reference it. “If posts are entirely deleted, it is probably due to the users themselves deleting posts.” I consider the habit of moderators deleting users posts sometimes a bit cowardly. Moderators will delete a post, give no reason, and the cowardly part … the user doesn’t even know who did it. This has happened to me more than once in the past. One other thing on this topic. I also consider it cowardly that a moderator uses the private messaging to attack a user. If they have something to say, say it in the forum, this practice should be stopped by Magix. OTOH maybe Magix agrees with and encourages it, who knows.
-
Videoon 7/29/2021
CPU working hard when nothing is happening
@tim-neighbors Have you tried using proxies, to see if the problem goes away? If it does then it’s probably not a thumbnail issue.
-
Videoon 7/29/2021
Vegas Pro 18 build 527 not fully using GPU
Yeah, I know, but focussing on my hobbies helps me through it. Indeed. Only way to go. I don’t have a use case for a GoPro, but i’d go for the Black if I had, my buddy is thinking of moving up to something like that, as you say, the stabilisation sets it apart. “not sure about Vegas.” I’m not sure if the “heal” option would work in VP because they are not from the same clip, maybe there's a script that does it. Vegasaur … Join Events (Unsplit) This command rejoins the selected adjacent events that have been previously split to multiple pieces. For the function to work, all selected events should be consecutive segments of the same media file (but not necessarily, you can also merge different media files, which can be very useful in multicamera mode). Several segments that are placed on different tracks and which refer to different media files can be selected and then unsplit in a single operation.
-
Videoon 7/29/2021
Vegas Pro 18 build 527 not fully using GPU
@Richard-Green By the sounds of it, your render times are the least of your worries, sorry to hear what you have to go through. Modern medicine is fantastic though, had 2xc and sorted. Thanks for the GoPro info. I guess you use something like Mp4joiner for putting things back together https://havecamerawilltravel.com/gopro/join-gopro-video-chapters/ RE: missing frames: Would it be of use to ungroup the audio and video, stretch the audio to match the video length then group again?
-
Videoon 7/28/2021
Vegas Pro 18 build 527 not fully using GPU
@Richard-Green I redid the Laptop and PC, the laptop came in at 6m:9s and the PC 6m:02s PC specs in signature. Perhaps the laptop was running cooler earlier, but no doubt it was as reported, 5:54s. No special VP settings, just stock. Anyway, my buddy has one of those Helmet cams, not a GOPRO, am sending on your sample for him to check out. Which GOPRO are you using, it's really lovely footage, very clear. I think he will upgrade to a GoPRO, but not on helmet. The PC was only switched on from cold, so that's as good as it gets for sure.
-
Videoon 7/28/2021
Vegas Pro 18 build 527 not fully using GPU
@Richard-Green Hi Richard, these are screenshots of render template used and TM while rendering. Although observing the preview monitor there is a start slow start aspect to the render, its better than some very spikey Nvenc renders I often see, maybe because it's only FHD. I did this all on my laptop using a laptop RTX 3080, I'll test on my PC and see how that goes.
-
Videoon 7/28/2021
Vegas Pro 18 build 527 not fully using GPU
@Richard-Green Results ... ----------- File sizes, data rates and render times ... --------------------------------------- Source GOPRO file ................. 3.73 GB ... 44.983 kbps ...................... Davinci Output file ................... 3.31 Gb ... 40.015 kbps ... Render time 2m:33s on Richards PC VFR2CFR ffmpeg output file ... 3.23 GB ... 39.078 kbps ... Render time 2m:10s Vegas Pro 18 output file .......... 3.28 GB ... 39.663 kbps ... Render time 5m:54s Render Quality Metrics, higher is better ... ---------------------------------------- Davinci Output file ................... PSNR 30.143705 .. SSIM 0.944624 (12.566809) .. VMAF 96.661926 VFR2CFR ffmpeg output file ... PSNR 34.732611 .. SSIM 0.938329 (12.099217) .. VMAF 91.093168 Vegas Pro 18 output file .......... PSNR 38.368225 .. SSIM 0.969091 (15.099089) .. VMAF 96.890182 The % column is an aggregate of all values to give an easy overall idea of quality.
-
Videoon 7/28/2021
Vegas Pro 18 build 527 not fully using GPU
@Richard-Green Indeed. I have done my best to have the output file sizes and data rates ~ match your davinci resolve clip. I am testing the rendered out quality of three clips, against the original, your Davinci clip, a VP18 clip and an ffmpeg clip. Is running now but may take a while. I am expecting there to be little difference, if so, then we have a good basis of comparison.
-
Videoon 7/28/2021
Vegas Pro 18 build 527 not fully using GPU
@Richard-Green Ok, thank you, have that. Just something to be going on with, using ffmpeg Nvenc encoding it takes about 2m:16s to render a 41MB/s file.
-
Videoon 7/28/2021
Vegas Pro 18 build 527 not fully using GPU
@Richard-Green Ok, i've already downloaded the source file.
-
Videoon 7/28/2021
Vegas Pro 18 build 527 not fully using GPU
@Richard-Green Hi Richard, the original clip is of no use to me without the rendered out Resolve clip + the time taken to render out the resolve clip. I cannot render out the resolve clip as I only have VP. If you can also supply the resolve clip via a download link then I can render out a similar clip from VP and supply the Render Quality Metrics. But I will be using a different machine than yours to do this. Ideally what would be still better is if you made available all three clips, the Original source clip, the Resolve clip and the VP clip.
-
Videoon 7/28/2021
Vegas Pro 18 build 527 not fully using GPU
@Richard-Green And you've confirmed that the quality output is the same? or are you content with eyeballing it? If so, then without taking quality into account and matching the output this whole thread is an Apples to Oranges comparison
-
Videoon 7/28/2021
Vegas Pro 18 build 527 not fully using GPU
@fifonik “A bit off-topic: Recently I was told that FFMpeg QSV h264 and FFMpeg CPU x264 produces about the same quality on the same bitrate.” Thats good news, thanks. I can confirm that with my testing. I don’t have a previous Qsv ffmpeg based test, only the VP MC one. I’ve updated my FHD table, via my signature with an extra item 20 QSV ffmpeg.
-
Videoon 7/27/2021
Version 18 updates
@bjtap1 Its a near impossibility that there will be another VP18 update. Why not make sure you have the previous download 482 ... https://www.vegascreativesoftware.info/us/forum/vegas-pro-18-update-4-build-482--127859/ then if you install the last update 527 https://www.vegascreativesoftware.info/us/forum/vegas-pro-18-update-5-build-527--129047/ you can still go back to 482 .
-
Videoon 7/27/2021
Playback issue Vegas 18
@Richvideo “It is only in Vegas that it is struggling.” Probably because there is no HW decoding support. Maybe VP19 will oblige?
-
Videoon 7/27/2021
Playback issue Vegas 18
@Richvideo "I noticed that this clip is not being decoded by RTX 2080 either but the CPU seems to be playing it back fairly smoothly." Same here, plays aok on Nvidia only laptop, no HW Decoding. it's actually 10 bit 422, IPB, but its AVC not Hevc. @LukasDr I downloaded your two new clips, both have HW decoding in VP. The 29.970 fps ALL-I clip plays at full fps, but as in previous testing the 59.940 fps IPB clip staggers, even with HW decoding. Both are Hevc. I think that the combination of Hevc, IPB (Inter-frame), no HW decoding, 10 bit, is really going to put any machine under pressure.
-
Videoon 7/27/2021
Playback issue Vegas 18
@LukasDr Thanks for those uploads, from my previous post I had already two of your clips that helped in testing. @Richvideo I suspect that VP could maybe do better in deciding Lukas's clips using Nvidia. Your clip doesn’t appear to be decoded at all using Nvidia. Can your camera output an All-I codec?
-
Videoon 7/27/2021
Vegas Pro 18 build 527 not fully using GPU
@Richard-Green Just a few more thoughts on doing the quality comparisons. If the quality PSNR SSIM and VMAF are pretty close then done deal. If theres a good bit of difference then heres what I would do. I would start outputting a slightly larger file size and data rate file (for the lower quality file) until I got it very close in SSIM etc to the other file. Then the increased file size and data rates can be used to give a rough % difference in quality, as can of course the initial % SSIM etc differences when the file sizes and data rates were very similar.
-
Videoon 7/27/2021
Vegas Pro 18 build 527 not fully using GPU
@Richard-Green Understood. https://www.vegascreativesoftware.info/us/forum/ffmetrics-yet-another-program-for-quality-metrics-calculation--122246/ And @wwaag's Render Quality Metrics here … https://tools4vegas.com/library/ “What I would like to know is why there is such a huge difference in rendering times using the same hardware” I think Resolve utilises GPU'S differently, more fully than VP. It can use more than one. If you do do the quality comparisons then maybe update here, it would be nice to know.
-
Videoon 7/27/2021
Vegas Pro 18 build 527 not fully using GPU
@Richard-Green I don’t want to get into an argument about this, mainly because I don’t know everything by far, about how the two different systems encode when using Nvenc. I did attempt, see my above post, to give you that answer by speculating that the TM readout may not be displaying the Asics output, only the Gpu, in VP. I do understand that you are happy with your conclusions so far, so don’t wish to look further.
JN-
Offline, Last login 7/30/2021, Member since 1/11/2020