Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[fitbrit]

I don't know where to start asking questions about this, so I will start here.
Can MC act as a WDM-like video processor using MadVR for video signals originating outside MC? This would either be from within the same PC, or from an external capture device.
For example, video input from a hypothetical HDMI 2.0 or 1.4 signal, say from a video capture device from another, or the same, gaming PC?

The MadVR Envy is a very expensive device at $10K, but it offers some ease-of-use advantages. I'm wondering whether we could achieve a similar result using our high-powered PCs, some effort and other hardware (which might also be pricey).

[mattkhan]

Yes though hardware compatibility is a challenge. For example I use a magewell device, works fine under Windows, works very badly in MC, works fine in mpc-be (with jriver wdm for audio).  Generally speaking I have found it a quite fragile thing but certainly it can be done though note that capturing multichannel audio is another level of difficulty (not sure a working card exists for this)

[fitbrit]

Great information.
In my mind, the audio would already have gone through an HDMI AVR or processor, and I'd only be feeding the live video stream for processing through MadVR in MC. The AVR or processor would also allow for a default AV synch offset for systemic lag introduced by all the devices in the chain.
For example one plays a 1080p video on a firestick, plugged into an AVR. The AVR handles the audio and passes the video onto a capture device, which then feeds MC for MadVR processing and finally the super upscaled 4K video signal is sent to the projector.
That's the dream. lol

[mattkhan]

Remember you need to strip hdcp too

[fitbrit]

Yes, I omitted that for simplicity. It's done quite easily with HDMI splitters and other off-the-shelf devices.

[RD James]

Whoa, that's crazy.
I had no idea you could use MPC-BE to process a capture card's input with madVR.
 
I had an Elgato CamLink 4K here to test that with and there seems to be no latency penalty for doing this either. It measures the same as other software: about 70ms on average (seems to increment 1ms every measurement, rolling around from 75ms to 65ms).
Unfortunately the NGU upscalers don't seem to work with live video. The video feed freezes when I use them, which is a shame as it looked really good when I was doing comparisons.
 
Curiously, G-Sync also works perfectly.
So with my monitor set to 100Hz, going full-screen in MPC-BE updates at 60Hz automatically (it is not changing the refresh rate to 60). This doesn't happen with video files.

[fitbrit]

Wow, RD James, you just tried it? Cool. I don't have the hardware to do it yet. A pity about the NGU though, as that kind of upscaling would be one of the draws of a solution like this.

[RD James]

Wow, RD James, you just tried it? Cool. I don't have the hardware to do it yet. A pity about the NGU though, as that kind of upscaling would be one of the draws of a solution like this.

Yeah, I have a setup here to capture video so it was easy to test.
I got the CamLink 4K a while ago because I have a Sony mirrorless camera with a nice lens which puts out good quality video, but has a 10 minute recording limit on the device.
It was never really built for video, and because there's no active cooling, recording on the device causes it to overheat; but if you record externally via HDMI it can run for hours.

Though it's sold as a device to hook up any HDMI-equipped camera and turn it into a webcam, I've found the CamLink useful for capturing other devices via HDMI too - which is how I was able to test this.
To follow up on the previous post, I've now tested the latency with G-Sync active in full-screen mode, and that dropped it to 55ms on average rather than 70ms - which is about what you'd expect (reducing latency by one frame at 60Hz).
Depending on what you plan on using the device for, the CamLink may not be what I'd recommend though - as it's quite limited in some ways. I've found it to be convenient though, as it's a small and portable USB3 device not much bigger than a thumbdrive.

[fitbrit]

I looking at using some capture device that can be housed within a single half-height PCI bracket or even a full height one inside an HD-Plex H5 case, but without needing the PCIe bus. Just thinking about new products to release down the line and the ways my customers might want to use MC. Fewer and fewer are buying video disks and wanting to rip them, but more and more are using their AppleTVs and Rokus etc. to stream video. If the video quality of those could be improved in MadVR before sending to the display device, I think the HTPC could be a poor-man's video processor.

[mattkhan]

I use ngu in this mode without problems

[fitbrit]

I use ngu in this mode without problems

Very interesting. Do you know of any step-by-step guides anywhere to achieve what you're doing?

[mattkhan]

Very interesting. Do you know of any step-by-step guides anywhere to achieve what you're doing?

I haven't seen one. It is quite simple though albeit a small bit of scripting is required to make it user friendly (in my setup at least). I can add detail later on if you like.

[Hendrik]

Since this is about increasing quality, it should be noted that if you want to consider 4K and HDR, you'll quickly exceed the available bandwidth of such devices - especially over USB - and "lossless" quickly turns into "lossy".
Even PCIe devices like the Elgato 4K60 Pro only allow 4K60 with subsampling, not full 4:4:4/RGB - and thats over PCIe, which is multitudes faster then USB3.0

If there is any lossy compression step (ie. in H.264 or such), even if its high bandwidth, or even worse added chroma subsampling, it'll negate any advantages you might get otherwise.

[mattkhan]

I think that sort of source content implies hdcp2.2 doesn't it? I haven't come across a way to strip that so this approach, as far as I can see, is for HD sources.

fwiw https://www.magewell.com/tech-specs/pro-capture-hdmi-4k-plus is one device that, from the specs, would appear to handle this sort of bandwidth. At least I can't see anything that mentions it (whereas some of their other products do make it clear). That featur e doesn't come cheap mind you

[RD James]

Since this is about increasing quality, it should be noted that if you want to consider 4K and HDR, you'll quickly exceed the available bandwidth of such devices - especially over USB - and "lossless" quickly turns into "lossy".
Even PCIe devices like the Elgato 4K60 Pro only allow 4K60 with subsampling, not full 4:4:4/RGB - and thats over PCIe, which is multitudes faster then USB3.0

If there is any lossy compression step (ie. in H.264 or such), even if its high bandwidth, or even worse added chroma subsampling, it'll negate any advantages you might get otherwise.

Subsampled chroma is not really an issue if you're dealing with video, since video is all encoded in 4:2:0 anyway.
That could mean it is upsampled by the player, downsampled by the capture device, and then upsampled by madVR; but I think any negative effects would be minor.
 
The larger issue, I suspect, is that most of these streaming devices will be upsampling content to the device's output resolution rather than switching to a native output for the source; e.g. outputting 720p content at 4K rather than outputting a 720p signal. So you're going to miss out on most of the benefits of madVR's upscaling.
And some capture cards -like the CamLink 4K I'm using- either require a device reset or manually setting the input resolution every time the source changes.
The biggest issue I've run into with streaming devices is that none of them do a great job with DVD sources - either with poor upscaling or deinterlacing. And none do IVTC to 24p, as far as I'm aware.
 
For me, this discovery mostly works out as a lower latency and higher quality option to play Switch games on my PC via a capture card than going through software like OBS, if I'm not recording/streaming.

[fitbrit]

I think that sort of source content implies hdcp2.2 doesn't it? I haven't come across a way to strip that so this approach, as far as I can see, is for HD sources.

fwiw https://www.magewell.com/tech-specs/pro-capture-hdmi-4k-plus is one device that, from the specs, would appear to handle this sort of bandwidth. At least I can't see anything that mentions it (whereas some of their other products do make it clear). That featur e doesn't come cheap mind you

If that card would work for the envisaged purpose, the price is totally acceptable. An Amazon.ca link for me: https://www.amazon.ca/Magewell-Capture-HDMI-Plus-Video/dp/B077N9W5MB
And possibly this? https://www.amazon.ca/AVerMedia-GC573-Live-Gamer-4K/dp/B07DHSZC4K/ref=pd_sim_23_1/146-4602991-6151831

[fitbrit]

Since this is about increasing quality, it should be noted that if you want to consider 4K and HDR, you'll quickly exceed the available bandwidth of such devices - especially over USB - and "lossless" quickly turns into "lossy".
Even PCIe devices like the Elgato 4K60 Pro only allow 4K60 with subsampling, not full 4:4:4/RGB - and thats over PCIe, which is multitudes faster then USB3.0

If there is any lossy compression step (ie. in H.264 or such), even if its high bandwidth, or even worse added chroma subsampling, it'll negate any advantages you might get otherwise.

Does that mean this is a pipe dream, and would only be possible buying a dedicated processor like the Envy?
My customers would buy such a device; cost is not the issue. However, it seems a waste if they have an RTX2080Ti available to them in their media PC already. As a total n00b here in this realm, I was envisioning the concept as just requiring MC to somehow accept an HDMI-in signal. I know that the devices that allow this might be very expensive, but again, the cost is not that huge a barrier - within reason.

[mattkhan]

As far as I can see, the main reason to use such a beefy card is for HDR but there is no way (that I know of) to strip hdcp from uhd so whether the input card can handle it or not (and the magewell card seems like it can) is a moot point. For non UHD content, there is no problem if using a suitable capture card. For example the magewell device I use has an app (https://www.magewell.com/usb-capture-utility-v3) which lets you configure exactly what should be output and what should be processed by the FPGA onboard.

[RD James]

As far as I can see, the main reason to use such a beefy card is for HDR but there is no way (that I know of) to strip hdcp from uhd so whether the input card can handle it or not (and the magewell card seems like it can) is a moot point.

I'm pretty sure there are lots of inexpensive splitters that will bypass HDCP.
The first result searching for "bypass hdcp 2.2" brings up a box on Amazon that claims to do it for about $60, with reviews saying it works.

[mattkhan]

The first result searching for "bypass hdcp 2.2" brings up a box on Amazon that claims to do it for about $60, with reviews saying it works.

When I have looked at this in the past, every device had comments saying it either didn't work in one form or another. I just did the same search and found similarly caveated products. Do you have a link to one that is meant to work?

[fitbrit]

I haven't seen one. It is quite simple though albeit a small bit of scripting is required to make it user friendly (in my setup at least). I can add detail later on if you like.

Thanks very much. I may well take you up on that detail in the near future.

Here's an example of a device that seems to do the HDCP 2.2 stripping:
https://www.amazon.com/OREI-UltraHD-HDMI-Splitter-Port/dp/B07J2Q9NPH/ref=sr_1_3?keywords=1x2+hdmi+splitter+hdcp+2.2+ultra+hd&linkCode=sl2&linkId=058722523f0c78d7b9e5172939f04d4a&qid=1580281989&sr=8-3

[fitbrit]

I just recalled that I actually bought this device a few weeks ago for a different application that I never got around to trying. Now to to find that thing... lol

[mattkhan]

The questions on that linked device suggest it doesn't strip, would be interesting to know if it does though

[fitbrit]

The questions on that linked device suggest it doesn't strip, would be interesting to know if it does though

I got to this device by way of a video demonstrating HDCP 2.2 stripping by a member of this device's family. Said device may have been limited to 10 Gbps, but it was stripping 2.2, I believe. I guess that doesn't mean that this unit will do so, but I remain hopeful.