Update, Cont'd: (ACES: v1, Pixel format: 32-bit (full range)

Changed Project Properties to:
- Pixel format: 32-bit floating point (full range)
- ACES version: 1.0
- ACES color space: Default (ACES2065-1)
- Best rendering quality (setting as before)
- Audio: Sample rate: 96Khz, 24 bit depth, Best resample

- Project timeline playback specs after these changes:
- CPU utilization now shot up to between 54% to 61%
- GPU 0: 7%, GPU 1: 3%

The same clips on this timeline played back just as smoothly as before (with the same two brief image hesitations mentioned previously)

New render status times:

- CPU: 29% @ 4.13Ghz

- GPU 0: 13%, GPU 1: 6%
- 27%, elapsed time 20:30, 53:50 left approx.
- final render time for same project file: 1:13:52 (@ 8 bit this render time was 13m 54s)

- The rendered file played back smoothly at UHD full screen (with no playback hesitation as in the preview with those two isolate spots so I should have first defragged that project file SSD first).

Wolfgang, I now see from your project task description what type of playback performance specs you're needing to know and I'll try to set up a test render with those particular parameters in mind. Sorry, but I'll first have to learn more about this software's capabilities before I can try and do that kind of test for you. Please feel free to guide me with some instructions if you'd like so I can get that done more quickly. Thanks!

Thank you, that is great. Depends what type of footage you have available. For me UHD 50p vlog or slog would make most sense. With a color transformation in the project settings to rec709.

@MelvinGonsalvez Are you saying a i9-9900x can playback a 4K 50p file in a 32bit float project with ACES at Best/Full at full frame rate?

My Sony AX100 that came out a few years ago is a 4K camera but it can only record 30fps at that frame size. I've since bought Sony's AX53 which also only does 4K recordings at 30fps max. with stereo audio but I choose the FHD setting which then allows me 60fps and 5.1 audio recordings at 1920x1080, 60p.

My test 4K project timeline I played back smoothly in 32bit FP (at the viewport size mentioned) had several AX100 AVCHD clips in a serial layout but they're only recorded at 30fps due to that specific camera's limited capabilities.

My Sony AX100 that came out a few years ago is a 4K camera but it can only record 30fps at that frame size. I've since bought Sony's AX53 which also only does 4K recordings at 30fps max. with stereo audio but I choose the FHD setting which then allows me 60fps and 5.1 audio recordings at 1920x1080, 60p.

My test 4K project timeline I played back smoothly in 32bit FP (at the viewport size mentioned) had several AX100 AVCHD clips in a serial layout but they're only recorded at 30fps due to that specific camera's limited capabilities.

@MelvinGonsalvez I am still not clear. So you used a 1080p AVCHD file in a 4K 32bit project for testing? FYI, I used to own the AX100 but upgraded to the AX700 last year.

I just build a i7-9800x system and it definitely cant not playback 4K 30p in 32bit float project at a reasonable fps at Best/Full. In fact, I would could call it unusable and I dont see how a i9-9900x would do any better.

No OldSmoke, I mentioned that my 4K testing was entirely with UHD recordings off my Sony AX100 at 30fps in stereo.

I was just reading up on the AX700 and am tempted by it but I'm not sure offhand how much better the recorded visual quality of the AX700 is in comparison. I've checked out some AX700 recordings online in the meantime. What's your opinion of this difference that you're now experiencing between the two systems?

No OldSmoke, I mentioned that my 4K testing was entirely with UHD recordings off my Sony AX100 at 30fps in stereo.

I was just reading up on the AX700 and am tempted by it but I'm not sure offhand how much better the recorded visual quality of the AX700 is in comparison. I've checked out some AX700 recordings online in the meantime. What's your opinion of this difference that you're now experiencing between the two systems?

The AX100 is a great camera and I had it for over 3 years. In comparison, the AX700 has picture profiles and matches my a6300 much closer. The image is a bit better too thanks to the RS sensor. Another good feature is the dual SD card slots and full size HDMI output.

Coming back to the i9-9900x system. I think I finally got it right. You used the AX100’s 4K XAVC-S 100Mbs files in a 32bit float 4K 30p project and you get 30fps playback with the preview at Best/Full?

For the testing it is not so important if you use UHD 30p or UHD 50p. Similar to OldSmoke I would expect to see a significant drop in the playback behaviour in fps, if you apply an ACES workflow that includes also a Color Transformation (do you know how you enable that? If not please say so).

That is why I tend to work with preview/quaterly in such a Project - and even with my older 8 core system overclocked and an R9 390X I do not exceed more then 5-7 fps wich such an ACES project.

Typically you can improve the playback behaviour by adjusting the preview ram (you have to test on your specicif system what tends to be the best number of mb you allocate). And with a good GPU too.

And: I also doubt a little bit if you are able to bring up the fps in such an ACES Project up to the 30 fps really, with the actual version of Vegas. Even not with 10 or 12 or 18 cores.

No OldSmoke, I mentioned that my 4K testing was entirely with UHD recordings off my Sony AX100 at 30fps in stereo.

I was confused on that point as well initially. For me, this is the most useful bit of info. If you can play out of camera UHD at full framerate in 32 bit mode, that's a substantial performance increase over my i7 4790K system. Out of camera footage from my A6500 in 32 bit ACES can only manage about 7fps. I guess the upgrade is in the cards after all. I only wish it was in the budget. ;-)

Thanks for the all the info Melvin.

Please feel free guys to double check my findings as you all are a lot more familiar with VP than I am currently and I don't want to inadvertently steer you the wrong way with any possible mistakes I've made on my end with reporting these initial tests.

My Project Properties on this test UHD build is showing as follows:

VIDEO TAB:

Template: Custom (3840x2160, 29.970 fps)
Width: 3840, Height: 2160
Field Order: None (progressive scan)
Pixel Aspect Ratio: 1.0 (Square)
Output Rotation: 0 degrees (original)
Frame Rate: 29.970 (NTSC)

Stereoscopic 3d Mode: Off

Pixel Format: 32-bit floating point (full range)
Compositing Gamma: 1.0 (Linear)
ACES Version: 1.0  (Set at this value & I don't know offhand what the other choice of 0.7 does.)
ACES Color Space: Default (ACES2065-1).  (I don't know exactly what the other four choices do.)
View Transform: Off  (I don't know what all the other drop-down choices do offhand.)
Full-Resolution Rendering Quality: Best
Motion Blur Type: Gaussian (I don't even know what's the best quality choice here.)
Deinterlace Method: None
Resample Mode: Disable resample.

AUDIO TAB:

Master Bus Mode: Stereo
Number of Stereo Busses: 0  (I didn't play with this setting even though I was tempted to put a value of "1" there.)
Sample Rate (Hz): 96,000
Bit Depth: 24
Resample and Stretch Quality: Best

PREFERENCES: (Video Tab)

Dynamic RAM Preview Max (MB): 200 (I'll next try bumping this way up to 2000+ to see the difference.  I currently only have 32GB system memory installed but I was planning to bump that up to 64GB at some point.)

Max. Number of Rendering Threads: 4  (I started with that default number that but was planning on upping it to 24 and see how that does as that's the maximum number this hyperthreaded i9-9920X has.)

GPU Acceleration of Video Processing: NVIDIA Corp. (GeForce RTX 2080 Ti).  I have two of these cards installed in SLI with nVLink but I'm not sure how well VP16 even supports this now.)

PREFERENCES: (Preview Device Tab)

Optimize GPU display performance: Enabled

PREFERENCES: (Internal Tab)

I didn't change any of the default settings in this Internal settings area. But I scrolled through and noted the following:

Maximum Video Render Threads (64-bit): 4 (As that was the default setting in the Video Tab.)
Disable Multi-Core Rendering: FALSE
Enable Multi-Core Rendering For Playback: TRUE
RAM Preview Limit (MB): 1,024
RAM Preview Limit (64-bit) (MB): 32,768

Please advise if you note any performance setting changes to be made.

@MelvinGonsalvez View Transform makes all the difference. When you enable any of the profiles, you will see the preview performance go way down especially with preview set at Best/Full... I am still missing the information from you.

My i7-9800x like your i9-9920x handles 32bit full without any issue as long as. jew Transform is off.

OldSmoke, I changed my Project Properties View Transform setting to Rec. 709 (ACES) from "Off" as per Wolfgang's suggestion on 7/17 and played back the timeline with the following results:

CPU: 55% (fluctuating around there)
GPU 0: 8%, GPU 1: 3% (fluctuating around those values)
Viewport Playback: Best (Auto)

The overall viewport playback (with the color shifted imagery) speed of the clips was quite smooth as before and the viewport settings were as follows:
Project: 3840x2160x128, 29.970p
Preview: 960x540x128, 29.970p
Display: 924x520x32 Rec.709 (ACES)

Viewport at Best (Full): ~ 7 fps

Viewport at Best (Half): ~ 12-15 fps

Viewport at Best (Quarter): Seems like real-time

View Transform Set To P3-D60 ST2084 4000 nits (ACES):
Playback of timeline was smooth as before.
Render Interim Results:
CPU: 50% (fluctuating)
GPU 0: 11%, GPU 1: 8% (fluctuating)

@MelvinGonsalvez "Viewport at Best (Full): ~ 7 fps" and that is the important information... no difference to my i7-9800x.

I'll try to find out what exactly my two RTX cards are doing for the project to hit even those low usage specs during playback and rendering. I'm not sure offhand if it's some type of image preprocessing, etc., tasked to them but there sure is a lot of GPU (and CPU) performance headroom for VP programmers to harness if more program-specific multi-threaded CPU/GPU code can be introduced into future builds.

Other graphics programs running on this system utilize a lot more CPU and GPU cycles when running physics and real-time interactive graphics simulations, etc.

I'll try to find out what exactly my two RTX cards are doing for the project to hit even those low usage specs during playback and rendering. I'm not sure offhand if it's some type of image preprocessing, etc., tasked to them but there sure is a lot of GPU (and CPU) performance headroom for VP programmers to harness if more program-specific multi-threaded CPU/GPU code can be introduced into future builds.

Other graphics programs running on this system utilize a lot more CPU and GPU cycles when running physics and real-time interactive graphics simulations, etc.

My understanding is that video encoding is very different and doesn't benefit from multiple GPUs. As of today, Vegas does not support SLI, CrossFire or NV Link and not even DUAL GPUs. BUT, you could use an AMD card for FX processing and an NVidia card for rendering with NVENC or CUDA depending on the model of NVidia card. Similar works for those with an Intel CPU that support QSV.

BUT, ...... and an NVidia card for rendering with CUDA.

Only old card and old Vegas versions can work with a little bit CUDA and most of the time is pretty faulty. Modern cards and the 2 last Vegas versions work with NVENC and that is something else as CUDA

BUT, ...... and an NVidia card for rendering with CUDA.

Only old card and old Vegas versions can work with a little bit CUDA and most of the time is pretty faulty. Modern cards and the 2 last Vegas versions work with NVENC and that is something else as CUDA

 

Corrected accordingly.

Viewport at Best (Full): ~ 7 fps

Viewport at Best (Half): ~ 12-15 fps

Viewport at Best (Quarter): Seems like real-time

Please try to Change that to preview/full, preview/half, preview/quater. That should bring up the fps significant.

And - also very important -what type of footage do you use? From with camera?

@Wolfgang S. You know, you could save yourself a helluva lot of back and forth by making up a small sample project, to your taste, and let @MelvinGonsalvez and myself and anyone else with a i9 9900n test it to your requirements. I made this offer earlier to @JoeAustin.

Ok, at this stage, maybe not worth your while as your probably nearly there.

I didn’t offer to comment, put together something else earlier in the thread simply because this type of project is all above my competancy level, but testing and giving playback rates etc I can do of course.

Viewport at Best (Full): ~ 7 fps

Viewport at Best (Half): ~ 12-15 fps

Viewport at Best (Quarter): Seems like real-time

Please try to Change that to preview/full, preview/half, preview/quater. That should bring up the fps significant.

And - also very important -what type of footage do you use? From with camera?

@Wolfgang S. From his posts I read that he is using his AX100 footage which is 4K 30p XAVC-S 100Mbps 420 8bit.

Ok, I had not seen that. Thank you.

@Wolfgang S. You know, you could save yourself a helluva lot of back and forth by making up a small sample project, to your taste, and let @MelvinGonsalvez and myself and anyone else with a i9 9900n test it to your requirements. I made this offer earlier to @JoeAustin.

Ok, at this stage, maybe not worth your while as your probably nearly there.

I didn’t offer to comment, put together something else earlier in the thread simply because this type of project is all above my competancy level, but testing and giving playback rates etc I can do of course.

Wright. That would be the best approach for sure. Unfortunately I have here a lot of different footages, what makes that not so easy. XAVC I slog 3 from my FS7. ProRes from my Shogun. Mp4 v log from an EVA1, all-i but also long GIP. Oh, also some iphone footage. That makes the exercise a little bit harder.