Datapath Vision E1 help needed:

[Fudoh]

It's great, yes. Recently I had a mainboard that just wouldn't run it (Intel H110 chipset), but I extensively tried the card on other mainboairds where it performed fine.

[paulb_nl]

Does anyone have stuttering issues with 1080p input? I use the 240p test suite scroll test and I see a lot of stuttering in the preview window of OBS and the Vision application with 1080p input and resolution set to 1920x1080. Surprisingly there is no stuttering when setting the cards resolution to 1920x1200 or even 1920x1081.

[SavagePencil]

I’ve seen conflicting reports about whether an upscaler offers any advantage with this card (versus going straight in with analog). What’s the benefit for having an FM or OSSC in the mix?

[FBX]

I’ve seen conflicting reports about whether an upscaler offers any advantage with this card (versus going straight in with analog). What’s the benefit for having an FM or OSSC in the mix?

I'd say at least in the OSSC's case, you get to use optimal timing mode to scale the image 'digitally' where pixel edges are razor sharp. Keep in mind the way optimal timing works is it samples the color and then represents the value for each pixel sample as a digital square. Thus it has the effect of making an analog console image look like it came from an emulator.

[SavagePencil]

Gotcha. I wasn’t sure exactly how good the digitization in the card was on its own, or how much 4:4:4 saved it. I was pretty impressed with the quality this guy was getting without an upscaler: http://speedrun.evilash25.com/2017/11/p ... h.html?m=1

Still have to admit your profiles, like the SMS one you did recently, are probably the cleanest I’ve seen!

[FBX]

I was pretty impressed with the quality this guy was getting without an upscaler: http://speedrun.evilash25.com/2017/11/p ... h.html?m=1

In that article, he admits to using post-processing to "point-resize" the images captures using nearest neighbor in order to upscale them. So he's effectively imitating what the OSSC does both automatically and more accurately (provided you dial in the optimal timing correctly). He's having to use Avisynth scripts to smooth out noise, which again, is covered by the OSSC's LPF function. So with the OSSC, you literally just feed say line4x optimally timed into the E1S, and then capture the output without having to worry about post processing.

[Fudoh]

It's also about usability and your workflow. If you capture native (low-res) you HAVE TO capture in 4:4:4, must do your initial post processing incl. scaling in 4:4:4 and MUST only encode into 4:2:0 (if you intend to use h264 for YT or anything similar).

If you upscale externally before capturing (ideally on a OSSC with a full 4:4:4 processing path), you can directly capture into 4:2:0 and you don't have to worry about 4:4:4 post processing to start with.

[SavagePencil]

If you’re saving it for archive, 4:4:4 would be best obviously. But I see your point.

[Fudoh]

If you're archiving low-res material, you still have to worry about proper playback all the time. Upscaled material (and stored in 4:2:0) is so much easier to handle.

[SavagePencil]

My point is that if I store it as low-res, I can preserve all the color data at 4:4:4 and with minimal storage. And then whenever we need to view that on a modern device, we can upscale to whatever we need with whatever color level we need.

[Fudoh]

yes, that's right, but it's not really easy access in nice quality, since you'll end up watching the clips with shitty live upscaling most of the times.