INTERACT FORUM

Please login or register.

Login with username, password and session length
Advanced search  
Pages: 1 ... 3 4 5 6 [7] 8   Go Down

Author Topic: JRVR Windows Testing  (Read 44022 times)

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #300 on: December 02, 2021, 05:58:57 pm »

Sounds fair, and it's not like these older iGPU's will ever be using anything more intensive than Bilinear anyway and those with a stronger GPU will be wanted to use a different scaler.  The only issue is that the default is Bilinear which may not be a good idea if it uses the old method if the green line issue is present with Bilinear.  Do you have a file that shows the issue and I'll run it up on the NUCs to see if it shows (that is if rolling back to V88 will use the old method)?
Logged
JRiver CEO Elect

murray

  • Citizen of the Universe
  • *****
  • Posts: 544
Re: JRVR Windows Testing
« Reply #301 on: December 02, 2021, 06:11:07 pm »

Is Jinc going to be the best upscaler JRVR will ever have?
Nothing like NGU is planned for the future?
When we have huge screens 145"+ NGU was still the best on madvr for qaulity....
Logged

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10729
Re: JRVR Windows Testing
« Reply #302 on: December 02, 2021, 06:56:08 pm »

More advanced scalers are planned for the future. NGU itself of course is specific to madVR, but there are alternatives, namely RAVU and FSRCNNX are in the running. RAVU is a medium-performance scaler, and FSRCNNX is a high-end quality scaler, which also needs a good GPU to run.
Logged
~ nevcairiel
~ Author of LAV Filters

murray

  • Citizen of the Universe
  • *****
  • Posts: 544
Re: JRVR Windows Testing
« Reply #303 on: December 02, 2021, 07:05:10 pm »

More advanced scalers are planned for the future. NGU itself of course is specific to madVR, but there are alternatives.

Ok thank you..... I just bought MC28 to try JRVR but really dont want to test it until it has something similar to NGU sharp. I used Jinc and higher on madvr before they developed NGU, so at this point in time I will just stick with madvr. I know NGU is limited to madvr, but for you to have something as high end as NGU, do you have to develop that or is there something similar off the shelf you can use soon?

Or are there limitaions with JRVR that will never allow high end scalers to be added.

Sorry for all the questions but I just dont want to waste time changing when NGU now is so fantastic.
Logged

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10729
Re: JRVR Windows Testing
« Reply #304 on: December 03, 2021, 02:58:45 am »

Do you have a file that shows the issue and I'll run it up on the NUCs to see if it shows (that is if rolling back to V88 will use the old method)?

It seems to happen on most UHD BDs for me, but not on NVIDIA since they seem to color the outside black. Its particularly obvious on FastBicubic and other bigger area scalers, but I can still measure some discoloration even on Bilinear, which isn't great (screenshot and color picker, the last line has a slight green hint, the one above does not). It was particularly easy to see on my 16:10 display I use on my testing rig, because the aspect ratio letterboxes the image, and thus the line moves up from the bottom border.
Logged
~ nevcairiel
~ Author of LAV Filters

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #305 on: December 03, 2021, 03:10:51 am »

OK - I'll have a look on the NUC driving an OLED, but I'll never see it on the other one as it is driving a PJ where the outside of the image is in the black velvet boarder (edit - this would only be true on 16:9 content which is my screen size).  Both of these NUCs are using Bilinear.
Logged
JRiver CEO Elect

lello

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 553
Re: JRVR Windows Testing
« Reply #306 on: December 03, 2021, 03:25:16 am »

FSRCNNX is a high-end quality scaler, which also needs a good GPU to run.

Is an RX580 considered a good GPU?
Logged

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #307 on: December 03, 2021, 03:25:51 am »

Rolled back the NUC7 to V87 and pixel peeping that bottom line (or two?) of the image is black (looks blacker than the letterbox).  It is unnoticeable but if it was green that would be another story. 
Logged
JRiver CEO Elect

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #308 on: December 03, 2021, 03:44:18 am »

Also did the test on the 3090 with FastBicubic, took a screen shot and zoomed it 800%.  I see some colour bleed into the letter box in parts of the image but the last line it is mostly black and completely unremarkable.  I'm also running Win11 with HDR ON and passthrough if that matters.  I've also only got nvidia and intel to test, but there is no issue I can see on these systems.
Logged
JRiver CEO Elect

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #309 on: December 03, 2021, 04:32:34 am »

I'm thinking a "Fast and Furious" option makes sense.  I get that quality is #1 but the other aim is to have JRVR run on as a wide a range of HW as possible including iGPU low powered devices.  An option going to do so would be great as V87 just works on the NUCs with no discernible issues. 
Logged
JRiver CEO Elect

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10729
Re: JRVR Windows Testing
« Reply #310 on: December 03, 2021, 05:53:02 am »

Is an RX580 considered a good GPU?

I wouldn't particularly call it very good, its getting a bit old afterall, but I can't judge how these will perform yet, we'll just have to wait and see (and try). There will likely also be multiple performance levels to pick from, and of course towards the top you have diminishing returns in quality as well.

I'm thinking a "Fast and Furious" option makes sense.  I get that quality is #1 but the other aim is to have JRVR run on as a wide a range of HW as possible including iGPU low powered devices.  An option going to do so would be great as V87 just works on the NUCs with no discernible issues.

I have changed LAV Video to color the frames entirely black before decoding to mitigate the impact from this, but until that rolls out I'm not really happy to offer an option that adds a potential green border (a slight darkening of the bottom border is far less observable, especially with most movies being letterboxed). So will just have to be patient for a while.
Logged
~ nevcairiel
~ Author of LAV Filters

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #311 on: December 03, 2021, 02:03:58 pm »

Nice idea.  There is always more than one way to skin a cat! Now to test SuperRes :)
Logged
JRiver CEO Elect

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #312 on: December 03, 2021, 02:33:53 pm »

Sent some Logs as V90 is sometimes crashing for me when testing Superres (not always but often when starting a video playback, MC just disappears). 

My first impressions is Superres certainly improved ringing artefacts on a broadcast TV eps. 

I also noticed that Ctrl-J does not work with Live TV even though JRVR is in the filter chain. 

With the dithering Algo's is Blue Noise the "Best" but most computationally expensive as I see in the notes that White is fast and noisy, and Ordered is balanced.
Logged
JRiver CEO Elect

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10729
Re: JRVR Windows Testing
« Reply #313 on: December 03, 2021, 03:10:21 pm »

Yes, Blue Noise is both the highest quality and most computationally expensive (if something would be better and cheaper, we would just drop the expensive one :D). Blue Noise is nice because it has reduced lower frequencies, which would be particularly distracting in dither noise. Ordered can show regular pattern which some people don't like, and White Noise is just a bit more noisy since it contains those low frequencies - but its fast.

I'm not sure how much of an impact it makes computationally, but since the option exists, might as well offer it.

PS:
The crash should be fixed in the next build. Thanks for the log.
Logged
~ nevcairiel
~ Author of LAV Filters

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #314 on: December 03, 2021, 04:25:45 pm »

Thanks for the info on Dithering. 

Is the crash just related to Superres or should I roll back to V88/89?
Logged
JRiver CEO Elect

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10729
Re: JRVR Windows Testing
« Reply #315 on: December 03, 2021, 04:57:00 pm »

Its just SuperRes related.
Logged
~ nevcairiel
~ Author of LAV Filters

Manfred

  • Citizen of the Universe
  • *****
  • Posts: 1023
Re: JRVR Windows Testing
« Reply #316 on: December 04, 2021, 06:58:22 am »

I have a new AMD 5700G APU and LG 34UC98-W PC-Monitor. With my old workstation I used only 8bit. The LG monitor allows also to use  10 bit = 8bit + A-FRC (Frame Rate Control) with leads  in the AMD display settings to RGB 4:4:4 Full RGB.

Is A-FRC good for JRVR or should I go with 8-bit?
Logged
WS (AMD Ryzen 7 5700G, 32 GB DDR4-3200, 2x2 TB SDD, LG 34UC98-W)-USB|ADI-2 DAC FS|Canton AM5 - File Server (i3-3.9 GHz, 16GB ECC DDR4-2400, 46 TB disk space) - Media Renderer (i3-3.8 GHz, 8GB DDR4-2133, GTX 960)-USB|Devialet D220 Pro|Audeze LCD 2|B&W 804S|LG 4K OLED )

bogdanbz

  • World Citizen
  • ***
  • Posts: 181
Re: JRVR Windows Testing
« Reply #317 on: December 04, 2021, 01:14:20 pm »

I noticed an issue with JRVR: it does not show external subtitles.

Test case 1:
- configure MC to use JRVR
- have a mkv or m2ts file with no internal subtitle
- have an external ass subtitle file
- start playback
-> the subtitle file is visible in the context menu, but is not displayed on screen

Test case 2:
- configure MC to use madVR
- have a mkv or m2ts file with no internal subtitle
- have an external ass subtitle file
- start playback
-> the subtitle file is visible in the context menu, and is displayed on screen

Test case 3:
- configure MC to use JRVR
- have a mkv or m2ts file with internal subtitles
- also have an external ass subtitle file
- start playback
-> all subtitles are visible in the context menu, with the external subtitle file being selected, but is not displayed over the video
- switch to an internal subtitle using the Subtitles context menu options, and then back to the external subtitle
-> now the external subtitle is displayed over the video !
Logged

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10729
Re: JRVR Windows Testing
« Reply #318 on: December 04, 2021, 03:13:17 pm »

The subtitle issue should be fixed for the next build.
Logged
~ nevcairiel
~ Author of LAV Filters

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #319 on: December 05, 2021, 04:26:50 am »

Thanks - SuperRes crash is fixed in V91.  Looks good.
Logged
JRiver CEO Elect

bogdanbz

  • World Citizen
  • ***
  • Posts: 181
Re: JRVR Windows Testing
« Reply #320 on: December 05, 2021, 04:47:32 am »

Thank you for fixing the external subtitle issue!
Logged

datdude

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 2215
Re: JRVR Windows Testing
« Reply #321 on: December 05, 2021, 04:35:37 pm »

The fix for the green line on 4:3 videos is gone and I haven't seen any lines for widescreen movies at the bottom.

SuperRes looks good and I haven't see any negative impact on performance.

Nice job all the way around!
Logged
"You are not a beautiful or unique snowflake." -  Just a very big snowball

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #322 on: December 06, 2021, 12:33:48 pm »

Looking forward to loading up the GPU's in V92 and testing the new scalers out.... but I'm away with work for a couple of days  :'( .  Have fun
Logged
JRiver CEO Elect

narbi

  • Junior Woodchuck
  • **
  • Posts: 69
Re: JRVR Windows Testing
« Reply #323 on: December 06, 2021, 12:59:43 pm »

My feedback on JRVR after using it since HDR passthrough is available.

I've been using JRiver with madvr for the past 5 years at least, I accepted some slight issues not that problematic, like stutter when JRiver OSD is active, things like that...

This is using a LG C8 as my only display, watching mostly 720p mkvs and ripped BD / UHD BD.

All I can say for now, is that it just works. Setup is standard, Jinc everywhere, HDR passthrough + windows switch.
I don't feel I'm missing anything quality wise from madvr with NGU sharp high everywhere + Jinc (if needed).
Picture is just as sharp, colors are just as right, HDR seems about the same.
The OSD doesn't cause stuttering anymore which is very nice.

Madvr seems abandoned a bit in favor of Envy (which is of course logical), it is better moving forward to bet on a new renderer which is actively developed and... just works.

The only slight issue I'm seeing, is sometimes when TV switches to HDR the HDMI goes nuts and the TV displays static noise. I have to power cycle the TV to get it working again. I don't remember having that issue with madvr, but at the same time I did upgrade to W11, and all the drivers with it, so who knows.

Keep up the good work.
Logged

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #324 on: December 07, 2021, 11:52:28 pm »

V93 Testing (on a 3090) with HDR Passthrough

WOW!
Quote
1. NEW: JRVR will show rendering performance metrics on the Info OSD (Ctrl-J).
Thanks - really helps in seeing what is going on.  As others have mentioned the ability to reset the stats would be good.

Quote
2. NEW: Chroma upscaling in JRVR can be enhanced by using Bilateral scaling (Chroma upscaling guided by Luma).
This works really really well and gives a very good quality bump even on my UHD HFR HDR material.  I know that many will say Chroma scaling is not that important (and hence the 4:2:0 encoding is used) but Bilateral scaling to bring it back to 4:4:4 is the best I've seen to date.

Quote
3. NEW: Image upscaling in JRVR can use advanced Image Doubling algorithms for enhanced quality when playing low resolution videos.
I tested on FSRCNNX16 across a range of low to high res material, and max the rendering time was only around 5ms with the 3090 so can't wait to see how the 1660Ti on the PJ goes when it gets darker.

One "issue" I did see was when the screen was not change refresh rate.  The first 1 or 2 seconds (say 100 frames @ 50hz) would have a weird "rush" effect until it settled down (I've attached the log).  It is hidden when change refresh rates (even with a delay on) but it looks a bit unsettling.  It may be worth having a bit of a wait for the queue (or an option for this?)?

Small one:  It would also be good to be able to get to the JRVR settings while the video is playing through Right Click --> Direct Show Filters --> JRVR (if possible) as it makes it easier for testing different settings as the image stays on the screen.
Logged
JRiver CEO Elect

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #325 on: December 08, 2021, 12:53:25 am »

Tested on a 1660Ti using HDR Passthrough to an OLED.

FSRCNNX16 can be too demanding of some material, but FSRCNNX8 was fine and the image looked great.
Logged
JRiver CEO Elect

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10729
Re: JRVR Windows Testing
« Reply #326 on: December 08, 2021, 02:53:37 am »

One "issue" I did see was when the screen was not change refresh rate.  The first 1 or 2 seconds (say 100 frames @ 50hz) would have a weird "rush" effect until it settled down (I've attached the log).  It is hidden when change refresh rates (even with a delay on) but it looks a bit unsettling.  It may be worth having a bit of a wait for the queue (or an option for this?)?

Waiting is not really feasible. madVR tried to do something like this and it just ended up rather buggy.
Its probably caused by compiling all the complex shaders, there is a shader cache planned that will help to improve startup time, but its quite a bit of work.
Logged
~ nevcairiel
~ Author of LAV Filters

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #327 on: December 08, 2021, 02:58:34 am »

Fair enough.
Logged
JRiver CEO Elect

voodoo5_6k

  • World Citizen
  • ***
  • Posts: 184
Re: JRVR Windows Testing
« Reply #328 on: December 08, 2021, 04:16:08 am »

With 28.0.92 available, I finally decided to also give it a try. Its main task is DVD upscaling (as >95% of my movies are on DVD, as *.iso on my server). With madVR, I had found a set of settings, organized in profiles, that would work very good with this SD content, while also taking care of the few BDs (also *.iso), and even fewer *.mkv and *.mp4.

I started with one of my worst case DVDs for upscaling (James Bond - The Spy Who Loved Me, original release, when there were still CRT TVs, no HDMI, but SCART etc.). And I must say, even with the highest scaling settings applied, it currently doesn't get close enough to madVR.

I have included two screenshots for comparison (cropped the black bars because of the file size limitations in the forum). Especially focus on the straight and high contrast lines on the Esprit's roof, windscreen frame, and door. Also, faces, Bond's right arm, and several other places. Overall, the picture doesn't look that natural and smooth, compared with madVR. In motion, it gets even worse. At first it looked awfully bad, but I found that, although I had selected OK after changing all the JRVR scaling settings, they were gone. I had to redo them, select Apply, and then OK, and finally, the high-end scaling selection persisted. Then, it was definitely better as before. But, currently, it's not working for me, DVD upscaling is too bad (compared to madVR), unfortunately.
Logged
END OF LINE.

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #329 on: December 08, 2021, 04:20:23 am »

Tested on a 1660Ti using HDR Passthrough to a 125" Screen (JVC x7500)

Looked good using Bilateral and FSRCNNX8. 

I really like HDR Passthrough to OLED/LCD and letting the displays doing their own tonemapping.  While I've currently got a pretty good image using HDR Passthrough with the JVC doing the tonemapping, I really need to spend more time using JRVR tonemapping as I get the feeling I'm leaving a lot on the table.  Hopefully, we will see more PJ owners testing!

Logged
JRiver CEO Elect

lello

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 553
Re: JRVR Windows Testing
« Reply #330 on: December 08, 2021, 04:46:30 am »

With my PJ I have tested BD and UHD with Bilateral, FSRCNNX16 and tonemapping and the result seems good to me provided I use standard resolutions. If I try with custom resolutions, my RX580 maxes out at RAVU.

I have to try the DVDs and the passtrough.
Logged

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #331 on: December 08, 2021, 05:03:37 am »

Thankfully, almost all my commercial videos are BD or UHD and the combo of Bilateral, and FSRCNNX works great for me.  I still have bunch of old camcorder footage (even Super8 conversions done years ago) and music video clips where the original quality is simply dreadful and my expectations for these are much lower. 
Logged
JRiver CEO Elect

voodoo5_6k

  • World Citizen
  • ***
  • Posts: 184
Re: JRVR Windows Testing
« Reply #332 on: December 08, 2021, 05:18:36 am »

Thankfully, almost all my commercial videos are BD or UHD and the combo of Bilateral, and FSRCNNX works great for me.  I still have bunch of old camcorder footage (even Super8 conversions done years ago) and music video clips where the original quality is simply dreadful and my expectations for these are much lower.
Yeah, that's an almost ideal case :) But for me, I'm not planning on replacing all these DVDs. Many of these aren't available on BD, and some are worse on BD than on DVD (massacred aspect ratio, oversaturated, too much contrast, odd colour tints etc.). So, I'm going to stick to these ;) Also, in many cases, the BD image is not what it looked like in cinemas. But that's personal preference of course.

I also have a few DVDs with even worse content (e.g. some NBA playoffs and finals from Bulls era), where no renderer achieved a really good result. Next time, I'll include these in the test suite too, I guess.

Oh, and I also don't have exaggerated expectations of course, for SD material. Matching madVR's results, that would be the moment, when I'll most likely be switching to JRVR permanently :) This time is not now, but JRVR is still so "young", and considering that, it has achieved a lot already :)
Logged
END OF LINE.

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10729
Re: JRVR Windows Testing
« Reply #333 on: December 08, 2021, 05:23:43 am »

The DVD results look more like a deinterlacing shortcoming to me, which is a current limitation - its not using hardware deinterlacing when you play through the DVD menu due to problems with the menu rendering.
This will be addressed through the subtitle re-working coming in the future (which also handles the menu), as well as a small update to LAV Video.

If you want to expirment a bit, you could rip one of those movies to an MKV and see if the quality improves, as that then won't have that limitation.
Logged
~ nevcairiel
~ Author of LAV Filters

voodoo5_6k

  • World Citizen
  • ***
  • Posts: 184
Re: JRVR Windows Testing
« Reply #334 on: December 08, 2021, 05:32:47 am »

The DVD results look more like a deinterlacing shortcoming to me, which is a current limitation - its not using hardware deinterlacing when you play through the DVD menu due to problems with the menu rendering.
This will be addressed through the subtitle re-working coming in the future (which also handles the menu), as well as a small update to LAV Video.

If you want to expirment a bit, you could rip one of those movies to an MKV and see if the quality improves, as that then won't have that limitation.
Interesting. I'll give the *mkv idea a try, either later today or tomorrow.
Logged
END OF LINE.

voodoo5_6k

  • World Citizen
  • ***
  • Posts: 184
Re: JRVR Windows Testing
« Reply #335 on: December 08, 2021, 10:14:07 am »

The DVD results look more like a deinterlacing shortcoming to me, which is a current limitation - its not using hardware deinterlacing when you play through the DVD menu due to problems with the menu rendering.
This will be addressed through the subtitle re-working coming in the future (which also handles the menu), as well as a small update to LAV Video.

If you want to expirment a bit, you could rip one of those movies to an MKV and see if the quality improves, as that then won't have that limitation.
Interesting. I'll give the *mkv idea a try, either later today or tomorrow.
I've just created a *.mkv from the DVD and was so curious I directly had to test it on that workstation :D It definitely looks better now! Way better! I'll make screenshots again, when I can test the *.mkv on the HTPC and compare to madVR (most likely tomorrow). Thanks, Hendrik :)
Logged
END OF LINE.

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #336 on: December 08, 2021, 02:13:30 pm »

Quote
Yeah, that's an almost ideal case :) But for me, I'm not planning on replacing all these DVDs.

I've fallen into the "trap" of now even replacing BD's for UHD's (about 25% of my library is now in that format) and no movie DVDs.... but I do still have music clips ripped from my DVD's so FWIW & to join the pixel peeping, here is an example of an MPEG file pulled from a DVD mastered Music Video that I've always had issues with, and then I picked a frame that looked as bad as possible.  On this content both JRVR and madVR show plenty of issues with combing (madVR is better) and random dots such as on Angus's top knuckle / tip of nose (JRVR is better).  This is why I've replaced what I can as I don't think there is any magic that will make this content look good.
Logged
JRiver CEO Elect

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #337 on: December 08, 2021, 03:36:40 pm »

On the 007 theme (left if madVR, right is JRVR same settings as above) and the source is 1920x1080 BD so both Luma and Chroma scaling (all non HDR). From what I can see (on the PNG Screen Shots - the attachment below is JPEG as the PNG is too big):
- Black crush in the jacket is handled better by JRVR
- Chroma scaling around the edge of the shirt looks better in JRVR
- Background noise is better in madVR (I can see some patterns in the noise with JRVR such as on the wall to the left of the head or the wall under the chin for example)

Overall this is really pixel peeping on still frames.  It looks good played in realtime.
Logged
JRiver CEO Elect

voodoo5_6k

  • World Citizen
  • ***
  • Posts: 184
Re: JRVR Windows Testing
« Reply #338 on: December 09, 2021, 01:04:24 am »

Overall this is really pixel peeping on still frames.  It looks good played in realtime.
Thanks for your comparison screenshots :) I'll add mine later, as soon as I've time to sit down in front of the TV. My example wasn't meant as pixel-peeping though, it did look even worse in motion. Wobbling lines, flickering edges (due to the aliasing), really bad. I agree 100%, the important part is what it looks like in motion, played at realtime speed. When one can't tell which version is madVR and which JRVR, then the image quality goal has been reached (for me) :)

This is why I've replaced what I can as I don't think there is any magic that will make this content look good.
The problem is, for a lot of content there are no better sources... And this highlights the biggest downfalls of our current display technology. Only a single resolution looks good. This has always bothered me once having to let go of CRT TVs (which have their own downfalls of course). This is a thing that has to be addressed via software, unfortunately, on lower resolution content, as the displays aren't capable of CRT like resolution flexibility. Unfortunately, this isn't a strong selling point (as the "consuming masses" always like "more", "higher" etc. better than having to actually understand the technology), otherwise we already would have something like that, instead of striving for ever higher resolutions without any content for it :D Also, that way they can sell the same content to the same person again and again, because that person "must" replace his/her BD with an UHD BD, and replace that with an HyperHD BD the next year ;) No offense, just joking and exaggerating :D
Logged
END OF LINE.

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #339 on: December 09, 2021, 01:52:31 am »

No offence at my end, and all very true.  Image the size and weight of a 65" CRT however! 

I too am not too keen on static comparo shots for the reasons you mention.... we don't watch video one frame at a time but in a flow of frames, and it's how the flow looks that is very important.  One argument I'd have for the rebuy is if the studio is doing its job competently (big if), then their ability to re-encode in non real time "should" be better than what we can do in real time at home.  Also, (and I'll be in the vast minority here) but I love the look of HFR, giving us additional temporal resolution running at 50 (or 60)fps (not just more pixels) and wish there were more directors like Ang Lee experimenting with this..... without the horror that is interlacing + it is fully compatible with all current displays and essentially a free doubling of (temporal) resolution.  I'm now shooting all my Home Videos @ UHD HDR 50fps with 360 degree shutter (eg no shutter which keeps the same cadence of "film / PAL" shot at 180 degrees but with half the motion blur) and it looks great, but of relevance to this thread is that JRVR's efficiency means for the first time even lowly HTPC's like my NUC7 and NUC8 can play these back without dropping frames.  This is something I could do with no other renderer. 

Looking forward to HyperHD BD!  Take my $$$$
Logged
JRiver CEO Elect

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10729
Re: JRVR Windows Testing
« Reply #340 on: December 09, 2021, 03:20:11 am »

I seriously doubt we'll see another incarnation of physical media unless some ground-breaking new technology emerges. Between 4K and 8K there are diminishing returns on the normal viewing distances at home, which makes it less desirable even for enthusiasts, never mind trying to market it to the masses. They are more likely to continue to do what they have done so far, and release re-masters of older content. Of course that might never reach every kind of niche content you might need to hang on to lower quality media for.
Logged
~ nevcairiel
~ Author of LAV Filters

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #341 on: December 09, 2021, 03:56:10 am »

The only thing we can trust with the studios is they will chase a $ wherever they can find it.  I've no idea if that includes new physical formats or not but they certainly shoot already on equipment capable of higher than 4k and likewise their production pipelines support higher resolution.  I'd personally like to see the back of low 24fps material but I think that is also very unlikely.  Anyway, for now we have plenty to work with on UHD content and as you point out the law of diminishing rewards is well and truly in play. 
Logged
JRiver CEO Elect

voodoo5_6k

  • World Citizen
  • ***
  • Posts: 184
Re: JRVR Windows Testing
« Reply #342 on: December 09, 2021, 06:04:28 am »

So, I did another check, now the DVD converted to *.mkv to remove the deinterlacing from the equation. Now, JRVR is relatively close to madVR in terms of image quality (in motion- as that's all that interests me). So, Hendrik was absolutely right. Once the menus and deinterlacing are fixed, this should work. It's still not where madVR is, but pretty close.

However, there are occurrences of scaling shortcomings which are more pronounced with JRVR. This time, look at the windscreen frame of the Esprit and its tailgate. In motion this is a flickering between relatively smooth and the aliased line you see on the screenshot. This also happens with madVR, but to a much lesser extend (I still notice it in motion, but it's not that distracting as it is with JRVR).

Overall, the picture is still smoother with madVR. But maybe, JRVR will be getting there, or even further... I can say, I'm impressed with what has already been achieved :) And although I'm not switching to JRVR right now, this might only be a matter of time. At least when the scaling for low resolution content will get better...
Logged
END OF LINE.

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10729
Re: JRVR Windows Testing
« Reply #343 on: December 09, 2021, 06:32:15 am »

Do you use quadrupling in madVR to go from SD to 4K? I suppose that could still help to some degree, we don't currently allow that.
Logged
~ nevcairiel
~ Author of LAV Filters

voodoo5_6k

  • World Citizen
  • ***
  • Posts: 184
Re: JRVR Windows Testing
« Reply #344 on: December 09, 2021, 09:48:49 am »

Do you use quadrupling in madVR to go from SD to 4K? I suppose that could still help to some degree, we don't currently allow that.
For SD, I'm using NGU Anti-Alias, with "high" for luma doubling and everything else on "let madVR decide".

Edit: Oh, and I don't use 4K. This is on a calibrated 1080p Panasonic WT50E. I'll only go higher in resolution when that TV dies some day. I don't need 4K or 8K for anything, so I'd refrain from buying something like that, but I want to calibrate my TV, so... I'll have to bite that bullet some day... Anyhow, I'm really happy with this TV, it's running for almost 10 years now, without any issues, and the calibration is the icing on the cake (for me).
Logged
END OF LINE.

armyplace

  • Recent member
  • *
  • Posts: 35
Re: JRVR Windows Testing
« Reply #345 on: December 09, 2021, 04:02:28 pm »

Ok it was a tough ask but I tried jrvr on a onboard hdmi ivy bridge intel cpu and for 4k material it's just too much. LOL

Playing 1080p remux movie though works very well and is quite smooth.

I'm going to install a gtx 1650 super gfx card in there and see how it performs with 4k material again.

Logged

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #346 on: December 09, 2021, 04:54:06 pm »

Yeah - I think the "bare minimum" iGPU for a HTPC these days is going to be Kaby Lake or later as this was when Intel added HW decoding of 264/265 UHD etc (so 7th Gen +).  You should have no issues once adding a discreet GPU like the 1650 Super however.
Logged
JRiver CEO Elect

armyplace

  • Recent member
  • *
  • Posts: 35
Re: JRVR Windows Testing
« Reply #347 on: December 10, 2021, 04:40:02 pm »

Yeah - I think the "bare minimum" iGPU for a HTPC these days is going to be Kaby Lake or later as this was when Intel added HW decoding of 264/265 UHD etc (so 7th Gen +).  You should have no issues once adding a discreet GPU like the 1650 Super however.

Yup, so last night I dropped in the gtx 1650 super and played some 4K movies with all scaling settings on maximum and judging by video, it looks buttery smooth. Very impressive. It's being used on a 1080p projector but I'll have a vava 4k UST projector coming in next week to see how it performs with JR DTM and also try HDR pass through.
Logged

BryanC

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 2558
Re: JRVR Windows Testing
« Reply #348 on: December 11, 2021, 06:25:41 pm »

According to the JRVR OSD stats my 4K rips (to 4K display) are being downscaled to 1080P? Is this correct?



I'm also having performance issues, may be related (unnecessary image downscaling)?
Logged

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14269
  • I won! I won!
Re: JRVR Windows Testing
« Reply #349 on: December 11, 2021, 07:19:10 pm »

Yup - It looks like your 4K display is set to 1920x1080
Logged
JRiver CEO Elect
Pages: 1 ... 3 4 5 6 [7] 8   Go Up