INTERACT FORUM

Please login or register.

Login with username, password and session length
Advanced search  
Pages: 1 [2] 3 4 5 6 ... 8   Go Down

Author Topic: JRVR Windows Testing  (Read 55347 times)

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10969
Re: JRVR Windows Testing
« Reply #50 on: October 15, 2021, 01:08:13 pm »

Yes, MC can reposition the video window any way you like, cutting off black bars etc. These settings are saved in the library, so its setup once (or setup externally). These settings are not available on Linux/Mac yet, but are implemented there, and will come in the next update. There is just no automation to get the right values - and of course you might have issues if you mix different screens.

There are potential issues with subtitles as I don't think it moves them into the visible section then, but subtitles are due for some reworking at some point.
Logged
~ nevcairiel
~ Author of LAV Filters

mattkhan

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 4271
Re: JRVR Windows Testing
« Reply #51 on: October 15, 2021, 01:44:49 pm »

Different screens would be a problem. Is it feasible to extend the mcws play command to take additional parameters to see such valurs on a per play basis? Or extend the library field to allow for per client settings?
Logged

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10969
Re: JRVR Windows Testing
« Reply #52 on: October 15, 2021, 02:41:37 pm »

I'm not in front of these right now, but I believe there are options to specify the cut-off for black bars only (ie. re-specifying the source rectangle), without forcing the image to change - which would allow the image to expand if your resolution allows it (eg. if the resolution is wider then 16:9), without causing issues on a normal 16:9 screen.

That is if your anamorphic setup actually uses a wider resolution in some fashion. If its just using image manipulation on a normal resolution that wouldn't work, but I have no experience with such setups.
Logged
~ nevcairiel
~ Author of LAV Filters

mattkhan

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 4271
Re: JRVR Windows Testing
« Reply #53 on: October 15, 2021, 02:57:28 pm »

In my case I use a 1.33x horizontal expansion lens which means zooming after the blacks bars are eliminated and then the lens stretches it back out to the correct AR. A more general solution will support different scaling ratios to support different lens.
Logged

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14497
  • I won! I won!
Re: JRVR Windows Testing
« Reply #54 on: October 15, 2021, 02:57:45 pm »

FYI I added Black Bar detection / crop factor to SOT - https://yabb.jriver.com/interact/index.php/topic,106802.msg904848.html#msg904848 
Logged
JRiver CEO Elect

lello

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 565
Re: JRVR Windows Testing
« Reply #55 on: October 15, 2021, 04:57:49 pm »

But you don't have build 76 yet.


If anyone wanted to test early, you can use a custom setup with LAV video and turn on D3D11.


It seemed to me that the test could be brought forward
Logged

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10969
Re: JRVR Windows Testing
« Reply #56 on: October 16, 2021, 01:57:25 am »

No, you can test with build 76 and the manual instructions, or an even later build when its all fully integrated. But build 76 is available publicly now, so you can do that now.
Logged
~ nevcairiel
~ Author of LAV Filters

mattkhan

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 4271
Re: JRVR Windows Testing
« Reply #57 on: October 16, 2021, 02:21:52 am »

I'm not in front of these right now, but I believe there are options to specify the cut-off for black bars only (ie. re-specifying the source rectangle), without forcing the image to change - which would allow the image to expand if your resolution allows it (eg. if the resolution is wider then 16:9), without causing issues on a normal 16:9 screen.
can you point me at the options for this pls? it's not obvious to me which fields control this.
Logged

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10969
Re: JRVR Windows Testing
« Reply #58 on: October 16, 2021, 02:40:12 am »

Right Click during Playback, the Window menu. If I take a typical cinema movie in 2.35:1 with letterboxing (eg. 16:9 in 1920x1080 or 3840x2160):

- Crop Black Bars -> 2.35
- Crop Black Bars -> Untick "Crop Sides of Video to Fill Screen"

This should just allow the video to fill the screen if there is room without any bad cropping or scaling.

In some tests I've had aspect ratio issues, but right now I can't reproduce them anymore. Otheriwse Override Aspect Ratio is there to resolve those.
This entire sizing logic is not new to JRVR, its been in MC a long time for DirectShow playback.
Logged
~ nevcairiel
~ Author of LAV Filters

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14497
  • I won! I won!
Re: JRVR Windows Testing
« Reply #59 on: October 16, 2021, 03:19:16 am »

...and from memory those settings are then saved into the "Playback Info" field and used for later playback.  You can also copy these settings to similar videos.
Logged
JRiver CEO Elect

lello

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 565
Re: JRVR Windows Testing
« Reply #60 on: October 16, 2021, 03:23:01 am »

Download v76 and do some short tests.

- zoom settings are now correctly saved.
- the tone mapping seems improved, but I need to verify better
- 60p files are played correctly, but a green strip has appeared at the bottom as indicated by jmone.Edit: only in case of custom setting, while if I let JRVR do everything automatically, no green strip appears, but the movie goes jerky as before

Thanks guys for the work you are doing
Logged

BryanC

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 2677
Re: JRVR Windows Testing
« Reply #61 on: October 16, 2021, 10:23:37 am »

So far, so good on my R1505G. I haven't thrown any super high bitrate stuff at it yet, but I've been using it to play 2160p h.265 hdr remuxes and 1080p h.264 files. I've noticed some microstuttering w/ certain files but I want to do some more testing before I can help diagnose.

Pretty amazing, this is a real game changer. When I have some more time I will start testing it out on Linux + Wayland.
Logged

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14497
  • I won! I won!
Re: JRVR Windows Testing
« Reply #62 on: October 19, 2021, 06:26:18 pm »

Hardware decoding should hopefully function without glitches now, and deinterlacing is done in hardware, both for software decode and D3D11 hardware decode (for what its worth, a feature madVR for some reason never bothered to implement, which is why D3D11 decode never saw widespread use there)

V77 Testing (yet to be posted on the main board) on the NUC7
- No need to use Custom mode anymore
- Green Line issue fixed
- Deinterlacing looks good, 1080p(i) --> UHD Scaling works nicely
- Can play (just) up to UHD HDR BD 23.976 with tonemapping, or UHD BD 50fps (not HDR / Tonemapping).  Can not play UHD HDR BD 50fps without dropping frames

So, it is pretty amazing that such a low powered and old iGPU can now handle a wide range of content.  One thing I would say is that you can not run anything else at the same time, even having Task Manager up (to see GPU / CPU utilisation) tips my NUC7 over the edge and I start to get dropped frames and out of sync audio on some material.

A job really well done and it will be interesting to see JRVR goes on newer NUCs etc.
Logged
JRiver CEO Elect

lello

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 565
Re: JRVR Windows Testing
« Reply #63 on: October 20, 2021, 03:22:55 am »

Rereading the latest posts, I had a question: does JRVR work on the GPU, as happens with madvr, or on the CPU or on both?
Logged

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10969
Re: JRVR Windows Testing
« Reply #64 on: October 20, 2021, 03:27:09 am »

No (sufficiently advanced) video renderer will ever work on the CPU, its just not possible. CPUs are not fast enough to do that.
Of course there is always some work thats done on the CPU, as is with everything, but no image processing or the likes should happen there.

On Windows JRVR uses D3D11 to use the GPU, on Linux/Mac it uses Vulkan or OpenGL. Technically it could also use Vulkan on Windows, but that would disable D3D11 decoding and deinterlacing features (at least for now), and there is no real major reason to switch either.
Logged
~ nevcairiel
~ Author of LAV Filters

lello

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 565
Re: JRVR Windows Testing
« Reply #65 on: October 20, 2021, 04:08:37 am »

Thanks for the explanations. So in the case of IdVR, it uses the integrated GPU, correct?
Logged

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10969
Re: JRVR Windows Testing
« Reply #66 on: October 20, 2021, 04:19:12 am »

Of course.
Logged
~ nevcairiel
~ Author of LAV Filters

lello

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 565
Re: JRVR Windows Testing
« Reply #67 on: October 20, 2021, 04:54:47 am »

Good! :)

So, given the current prices of the GPUs, it will be worthwhile to take the upcoming IdVR.
I wait to know all the specifics. For me it would be interesting, among other things, also that it had two HDMI outputs to connect one to the receiver (just audio would be enough) and the other to the pj.
Logged

tensionfire

  • Recent member
  • *
  • Posts: 8
Re: JRVR Windows Testing
« Reply #68 on: October 20, 2021, 03:43:55 pm »

This is an pretty cool new feature of JRiver. It looks good as long the brightness of the film is in a lower range. If you look at Films like the Shallows you have very big clipping problems.

I hope there will be a solution in the future for brighter films.
Logged

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14497
  • I won! I won!
Re: JRVR Windows Testing
« Reply #69 on: October 20, 2021, 04:08:46 pm »

I too see some SDR tone mapping issues on HDR material but keep in mind, it's only a preview so far.  Just the base platform is in and working.  All sorts of stuff still to be done, such as overlays (for BD Menus), HDR Passthrough (instead of HDR to SDR tone mapping), menu to control options etc etc etc.  Who know (well Hendrik does) how it will evolve over time. 

I'm pretty impressed how usable it already is.  I'm especially surprised at how well it does on low powered iGPUs and that the core is now cross platform.  I can certainly see JR switching the default renderer to JRVR on all versions of MC offering a good quality "just works" on a wide range of HW over EVR etc.  How it goes replacing madVR will depend on what features you need vs the JRVR dev path at any point in time.  At present I'm still using madVR but I get the feeling that may not last that long! 
Logged
JRiver CEO Elect

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10969
Re: JRVR Windows Testing
« Reply #70 on: October 21, 2021, 07:41:51 am »

The next build will add a first settings panel, so a few comments on some of those, specifically HDR tonemapping settings.

There are two tonemapping algorithms you can currently choose between, BT.2390 and Hable.
Both of these work much better with proper peak information, so either your file provides good metadata, or you use the Peak Detection option (although disabling that will offer good performance benefits on low-end systems).

And you can configure the target peak in nits to achieve a more appropriate tonemapping, although reducing the target from the default will amplify any short-comings in the tonemapping, so its not necessarily recommended. But ultimately this option should be set to taste, not some theoretical value.

Additionally, parameters for Desaturation are available. I won't go into the deep math about how they work, but here are two sets you can try:
- Default: 0.9 Strength, 0.2 Exponent.
- Alternative: 0.75 Strength, 1.5 Exponent

Of course you are free to try other numbers as well.

Depending on the content, a mixture of different desaturation settings and the algorithm choice can definitely help with tonemapping results that are too bright.

Other then tonemapping, scaling algorithms are selectable so far.  More settings to come.

On a more general note on the settings, the dialog has an "Apply" button which should apply the settings to the active video, so you can pause it, change settings, and hit Apply to see the changes. This works as long as the video is being shown in the active zone.
Logged
~ nevcairiel
~ Author of LAV Filters

tij

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 1563
Re: JRVR Windows Testing
« Reply #71 on: October 21, 2021, 11:38:42 am »

There are two tonemapping algorithms you can currently choose between, BT.2390 and Hable.
Both of these work much better with proper peak information, so either your file provides good metadata, or you use the Peak Detection option (although disabling that will offer good performance benefits on low-end systems).

I recall MadVR at some beta point had video analysed for brightness. It was ditched later in favour of Dynamic Tone Mapping. My guess it was no good for Envy.

But for MC this might be interesting. Maybe measure brightness values between chapters (takes care of complicated scene detection algorithms) ... UHD disc usually have pretty good chapters to separate "scenes" ... granted its not ideal (as chapter might have both bright and dark scenes) ... and its not on par with DTM in term of flexibility ... still beats a hell of a brightness value for whole movie
Logged
HTPC: Win11 Pro, MC: latest 31(64b), NV Driver: v425.31, CPU: i9-12900K, 32GB RAM, GeForce: 2080ti
Screen: LG 2016 E6
NAS: FreeNAS 11.1, SuperMicro SSG-5048R-E1CR36L, E5-1620v4, 64GB ECC RAM, 18xUltrastar He12-SAS3 drives, 2x240GB SSD (OS)

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14497
  • I won! I won!
Re: JRVR Windows Testing
« Reply #72 on: October 21, 2021, 03:38:22 pm »

First initial impressions (on a 3090 --> HDR Display) running MC V28.0.78
- Well laid out Menu
- Could we get the ability to access the JRVR Menu through Right Click --> Direct Show Filters --> JRVR like with madVR (as this way you still have Video Playback control, as going via Playback Settings you lose playback control)?
- I'm impressed with Hable over BT.2390 on my Max 1,000nit HDR BT2020/DCI P3 videos.  It does not blow out highlights (or make them iridescent) like 2390 (though is "darker" overall).
- I'll need to play more with Strength and Exponent but unless you put in crazy high strength #'s it seems fairly subtle
- Love that JINC is an option (no issues on a 3090) and will see what options work best on the NUC

Let the knob twiddling begin!
Logged
JRiver CEO Elect

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10969
Re: JRVR Windows Testing
« Reply #73 on: October 21, 2021, 04:30:13 pm »

BT.2390 definitely provides a brighter image overall, and with some adjustments to either peak nits and/or desaturation, you can also minimize the blow out. It's all a balancing act.

That said Hable looks pretty close to the madVR version we are shipping, in brightness too, so maybe that's a better option for now.

New algorithms might appear in the future, as the industry figures out what's best and research gets published.

On a quick side note, both Peak Detect and Jinc use compute shaders to achieve good performance, so performance might go down drastically when those are not supported - and i think Peak Detect just turns itself off.
Logged
~ nevcairiel
~ Author of LAV Filters

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14497
  • I won! I won!
Re: JRVR Windows Testing
« Reply #74 on: October 21, 2021, 04:42:06 pm »

These options will be great to "dial in" different displays, OLED Displays at 500ish Nits, Beamers at much lower Nits, and some LCD Screens at over 1,000 Nits...... or just use Passthrough to let a HDR Display do it's own thing. 

You could even make a "generic" selection option for those that don't know their displays' brightness
Logged
JRiver CEO Elect

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14497
  • I won! I won!
Re: JRVR Windows Testing
« Reply #75 on: October 21, 2021, 06:31:20 pm »

Did some more quick tests on the NUC (one to a LG OLED, the other to a Epson PJ in a dark room)
- I tried all combos of the current settings, these little PCs can not play back tone mapped HDR UHD 50/60fps material without dropping frames.  No issues with STD UHD BD however.  So unless HDR passthrough helps, it looks like HW upgrade time for me
- With out of the box settings, Hable's tone mapping works far better
- 400nits looks pretty good on the OLEDs with Hable and 200 on the PJ.

I'll have to wait till it is dark here in Oz to crank up the big JVC (with a 1060Ti) and see how that goes.
Logged
JRiver CEO Elect

JimH

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 72538
  • Where did I put my teeth?
Re: JRVR Windows Testing
« Reply #76 on: October 21, 2021, 06:58:00 pm »

Thanks again for all the testing and reporting.  Waiting for dark now.
Logged

tij

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 1563
Re: JRVR Windows Testing
« Reply #77 on: October 22, 2021, 01:27:34 am »

Setting Update on Beta Channel does not load 28.0.78. Have to manually load it.

Is there option to skip video by frame? Minimum i can set now is 5 sec. Nearly impossible to take screen shot of same frame for comparison.

First test Upscaling and Deinterlacing.

Source: US NTSC DVD of Futurama Season 1.

Observation:

Deinterlacing looks same to me in Video mode.

MadVR allows forcing Film mode which in my opinion is better for this particular DVD. Maybe JRVR can have some specific field/tag that allows user to control deinterlacing option (Auto, None, Film, Video)

Upscaling on MadVR is noticeably better for animated stuff ... not so noticeable for films

EDIT: these combing effects only pops in ocassionally ... my guess - bad editing (its even worse for PAL DVD) ... majority of stuff gets deinterlaced properly both in Film and Video mode ... like i mentioned, Film mode seems to hadle "bad" spots better for this DVD
Logged
HTPC: Win11 Pro, MC: latest 31(64b), NV Driver: v425.31, CPU: i9-12900K, 32GB RAM, GeForce: 2080ti
Screen: LG 2016 E6
NAS: FreeNAS 11.1, SuperMicro SSG-5048R-E1CR36L, E5-1620v4, 64GB ECC RAM, 18xUltrastar He12-SAS3 drives, 2x240GB SSD (OS)

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10969
Re: JRVR Windows Testing
« Reply #78 on: October 22, 2021, 02:37:56 am »

Upscaling on MadVR is noticeably better for animated stuff ... not so noticeable for films

Well you are using NGU, alternatives for that will be added at a later point. For now just normal scalers.

As for "film" deinterlacing, what that really means in madVR is that it actually turns deinterlacing off, and tries to do IVTC instead (which is entirely custom). But your file at hand is being detected as a 5:5 cadence, which is usually a sign of bad frame blending during mastering of the disc, or such shenanigans. I don't have any immediate plans for IVTC, but who knows in the future when everything else is done.

Of course that wont save that clip, even the film mode looks bad with the blending of the hand and bottle. If I cared about that content, I would try to offline process it and see if it can be rescued. Thats usually the best choice for old DVDs that don't have a more modern source.
Logged
~ nevcairiel
~ Author of LAV Filters

tij

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 1563
Re: JRVR Windows Testing
« Reply #79 on: October 22, 2021, 04:29:13 am »

Great to hear that alternative to NGU is on roadmap.

I always suspected those combing were mastering problem. IVTC for future would be great. Still have some old DVDs that will likely never see BD release.

Great progress on JRVR ... briefly looked at tone mapping and its great too ... will post observation on those later

PS. any way to skip video frame by frame in MC ... for easier screenshot takes
Logged
HTPC: Win11 Pro, MC: latest 31(64b), NV Driver: v425.31, CPU: i9-12900K, 32GB RAM, GeForce: 2080ti
Screen: LG 2016 E6
NAS: FreeNAS 11.1, SuperMicro SSG-5048R-E1CR36L, E5-1620v4, 64GB ECC RAM, 18xUltrastar He12-SAS3 drives, 2x240GB SSD (OS)

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14497
  • I won! I won!
Re: JRVR Windows Testing
« Reply #80 on: October 22, 2021, 05:35:11 am »

Did some testing and watching of content on my JVC X7500 (1660ti) and it looked great when the JVC was set to SMPTE ST 2084 (though it is a rather convoluted pipeline with JRVR tonemapping HDR to SDR --> Windows 11 then tonemapping SDR to HDR).  The result was every bit as good as what I was getting with madVR tonemapping.  Good details in the shadows and highlights well controlled.  Hable is still my preferred option.   I'm looking forward to seeing how HDR passthrough works, or even a SDR --> HDR tonemapping for those of us with HDR displays. 

Oddly, I did get periods where the video was dropping frames.  A pause/play would settle it down but it's not something I've seen for a long time so I'm unsure of the cause.  I bumped the NVidia profile to performance but that did not seem to help.
Logged
JRiver CEO Elect

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 42441
  • Shoes gone again!
Re: JRVR Windows Testing
« Reply #81 on: October 22, 2021, 07:14:53 am »

I just installed 77 again.  Then I checked for updates and build 78 was there.
Logged
Matt Ashland, JRiver Media Center

tij

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 1563
Re: JRVR Windows Testing
« Reply #82 on: October 22, 2021, 08:06:52 am »

I just installed 77 again.  Then I checked for updates and build 78 was there.
I was on 76 public release ... changed option to beta ...entered password ... and it said i was on latest
Logged
HTPC: Win11 Pro, MC: latest 31(64b), NV Driver: v425.31, CPU: i9-12900K, 32GB RAM, GeForce: 2080ti
Screen: LG 2016 E6
NAS: FreeNAS 11.1, SuperMicro SSG-5048R-E1CR36L, E5-1620v4, 64GB ECC RAM, 18xUltrastar He12-SAS3 drives, 2x240GB SSD (OS)

jkauff

  • World Citizen
  • ***
  • Posts: 202
Re: JRVR Windows Testing
« Reply #83 on: October 22, 2021, 09:38:43 am »

Glad to hear NGU is on the roadmap. I have lots of 480p movies, mostly black and white, carefully converted from DVD using Handbrake to de-comb, etc. Using madVR, I can get them looking almost as good as Blu-ray on my 4K TV.

JRVR, not surprisingly, doesn't even come close to madVR on 480p. I'm sure that's a low priority, except for people who want to play DVDs from their discs with a decent-looking picture.

I'll be staying with madVR on Windows, but I also have Mac and Linux machines that currently can't do better than VLC quality. it'll be great to have the option of JRVR, so don't forget us 480p folks.
Logged

lello

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 565
Re: JRVR Windows Testing
« Reply #84 on: October 22, 2021, 10:06:48 am »

From the beginning I had noticed that the image, especially of 4k HDR, was more compact, cleaner than madvr, and today I understood what it was: the almost total absence of video noise.

With madvr I had never been able to remove it, and in the end I thought, since in some films it was more present than in others, that it was a choice of the director, but now I have discovered that it is not so: does it depend on JRVR's best work? Or was it just me who didn't know how to use madvr (I've always used Asmodian's low settings)?
Logged

tij

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 1563
Re: JRVR Windows Testing
« Reply #85 on: October 22, 2021, 10:35:30 am »

From the beginning I had noticed that the image, especially of 4k HDR, was more compact, cleaner than madvr, and today I understood what it was: the almost total absence of video noise.

With madvr I had never been able to remove it, and in the end I thought, since in some films it was more present than in others, that it was a choice of the director, but now I have discovered that it is not so: does it depend on JRVR's best work? Or was it just me who didn't know how to use madvr (I've always used Asmodian's low settings)?

Screenshot comparison would be good ... to know exactly what you mean

In some case "film grain" is part of film "character" ... EDIT: and noise reduction (whether intentional or biproduct of other processing) can "flatten" out the image
Logged
HTPC: Win11 Pro, MC: latest 31(64b), NV Driver: v425.31, CPU: i9-12900K, 32GB RAM, GeForce: 2080ti
Screen: LG 2016 E6
NAS: FreeNAS 11.1, SuperMicro SSG-5048R-E1CR36L, E5-1620v4, 64GB ECC RAM, 18xUltrastar He12-SAS3 drives, 2x240GB SSD (OS)

tij

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 1563
Re: JRVR Windows Testing
« Reply #86 on: October 22, 2021, 12:22:54 pm »

OK ... did some testing with my OLED E6

I agree that Hable is close to MadVR. But I have some problems.

With MadVR i set target nits at 480. Doing 480 nits for Hable gives similar results where scenes are very bright, but dark scenes gets crushed.

Doing 100 nits for Hable gives similar to my MadVR, but then very bright scenes get "clipped".

See attached (used Batman v Superman for dark scene ... Harry Potter Hallows2 for bright scene)

Similar story with BT (didnt include screenshots ... afraid too overcrouded)

EDIT: desaturation settings were left at default
Logged
HTPC: Win11 Pro, MC: latest 31(64b), NV Driver: v425.31, CPU: i9-12900K, 32GB RAM, GeForce: 2080ti
Screen: LG 2016 E6
NAS: FreeNAS 11.1, SuperMicro SSG-5048R-E1CR36L, E5-1620v4, 64GB ECC RAM, 18xUltrastar He12-SAS3 drives, 2x240GB SSD (OS)

lello

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 565
Re: JRVR Windows Testing
« Reply #87 on: October 22, 2021, 12:48:31 pm »

In fact I should have done some comparison screenshots, it was the most logical thing, but I only trusted my eyes.

I then started MC with madvr to make the comparison (I hadn't used it anymore) and strangely the image didn't start (you could only hear the audio). I rebooted, and the image started but as soon as I changed the chapter everything froze: is it possible that JRVR and madvr went into conflict? Or is my htpc having problems?
Logged

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14497
  • I won! I won!
Re: JRVR Windows Testing
« Reply #88 on: October 22, 2021, 03:19:43 pm »

There will be no conflict between madVR and JRVR so it is something else?
Logged
JRiver CEO Elect

lello

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 565
Re: JRVR Windows Testing
« Reply #89 on: October 22, 2021, 03:57:58 pm »

I have no idea. Tomorrow I take a look at madvr settings and if it still doesn't work I try to reinstall it
Logged

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14497
  • I won! I won!
Re: JRVR Windows Testing
« Reply #90 on: October 22, 2021, 04:03:51 pm »

....well something weird has also happened to me with madVR on my main PC (just went to do comparison pics).  On Fury Road I was just getting mostly a black screen with the odd sec of visible video.  On other videos it was fine.  I also removed my custom madVR version and let MC dowload the shipped version and the reset the setting using MC --> Tools --> Options --> Quality Settings --> Best Quality.

This is one of the reason I'm keen to see JRVR Shine. 
Logged
JRiver CEO Elect

lello

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 565
Re: JRVR Windows Testing
« Reply #91 on: October 22, 2021, 04:06:47 pm »

What is JRVR shine?
Logged

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14497
  • I won! I won!
Re: JRVR Windows Testing
« Reply #92 on: October 22, 2021, 04:11:23 pm »

I agree that Hable is close to MadVR. But I have some problems.

With MadVR i set target nits at 480. Doing 480 nits for Hable gives similar results where scenes are very bright, but dark scenes gets crushed.

I agree that side by side the dark detail in JRVR/Hable is crushed vs madVR/ST2390 (the following is a comparison of both set to 750nits).  I also had a play with changing Strength and Exponent but changes are subtle unless you go wild.  Keep in mind however that there is no Target Gamut / Gamma options to set yet. 
Logged
JRiver CEO Elect

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14497
  • I won! I won!
Re: JRVR Windows Testing
« Reply #93 on: October 22, 2021, 04:14:32 pm »

What is JRVR shine?

As in do well.  I'd personally be very happy with JRVR being the renderer that "Just Works" and produces a great image across a range of equipment without the hair pulling you can get with madVR.  Don't get me wrong, madVR is very good but weird things like what I just got with Fury Road leaves your wondering what is going on. 
Logged
JRiver CEO Elect

JimH

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 72538
  • Where did I put my teeth?
Re: JRVR Windows Testing
« Reply #94 on: October 22, 2021, 06:17:18 pm »

What is JRVR shine?
It's a T-Shirt we have in development.  Might take a while.
Logged

JimH

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 72538
  • Where did I put my teeth?
Re: JRVR Windows Testing
« Reply #95 on: October 22, 2021, 06:18:43 pm »

I'd personally be very happy with JRVR being the renderer that "Just Works" and produces a great image across a range of equipment without the hair pulling ...
That's the goal.  Well said.
Logged

marko

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 9165
Re: JRVR Windows Testing
« Reply #96 on: October 22, 2021, 11:33:11 pm »

I'd personally be very happy with JRVR being the renderer that "Just Works" and produces a great image across a range of equipment without the hair pulling you can get with madVR.
This. I've read the entire thread, and, honestly, it's a foreign language to me. I see a lot of 'what's MC doing vs what's the TV doing' and such and often wonder if I'm getting the best out of the HTPC which is using the intel UHD 730 graphics. Everything is at default. MC uses ROHQ. When I switch the TV to the HTPC input, it shows an UHD symbol in the top right, and things look pretty impressive to me and my untrained, ignorant eyes!

I am certain that soon, I will reap the benefits of all this work and heavy testing and wanted to say a massive thank you to all involved in that time consuming process, which although hard to follow, is still fascinating to watch unfold.

-marko

mattkhan

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 4271
Re: JRVR Windows Testing
« Reply #97 on: October 23, 2021, 03:58:52 am »

I noticed the settings dialog is not saving settings per zone, i.e. change in one zone, switch the zone dropdown to another one, reopen the settings dialog, the settings are as per the previous zone. Not sure if intentional or just not implemented yet but thought I'd mention it in case it is meant to be per zone already.
Logged

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10969
Re: JRVR Windows Testing
« Reply #98 on: October 23, 2021, 04:04:50 am »

Its not meant to be per zone currently. I have not decided yet what to do about that, since once it gets down to it, people might also want more flexible options per-file or more generically depending on various file properties, and conversely zones might be for different audio devices (headphones vs speakers, or such) but for the same display, so its all not a great fit, and ZoneSwitch also doesn't have enough information to fully represent all video properties relevant for such decisions (which are only available during playback).

So I was considering a profile system driven by expressions to select the current one, or something like that, but thats for later.
Logged
~ nevcairiel
~ Author of LAV Filters

mattkhan

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 4271
Re: JRVR Windows Testing
« Reply #99 on: October 23, 2021, 04:21:38 am »

So I was considering a profile system driven by expressions to select the current one, or something like that, but thats for later.
that would be nice

fwiw it sounds like a similar problem as we have now with per file (audio) DSP configuration (where you often want to be able to use some base config and then customise certain aspects on a file basis), might be nice to solve this for video and then later apply that to DSP.
Logged
Pages: 1 [2] 3 4 5 6 ... 8   Go Up