Screen capture or screen capture through HDMI capture card?

[u666sa]

Hi. I want to be doing repair videos, I will be having 1 microscope camera, 1 overhead and 1 side camera, plus I want to capture screen to show boardviews and such. So question is which one is going to be better, capture screen through OBS or capturing HDMI signal (of the screen) through HDMI capture card?

My screen is 21:9, 1080p, 29 inch. I guess there is no real need to capture the entire wide screen, just a window. In case of PDF viewer it would be just one window. In case of FlexBV boardview software it would be 2 windows, 1 of the boardview and 1 of the PDF.

 

[AaronD]

A series of window captures might be best, as long as you don't close any of those windows.

Cheap HDMI captures (~$20 USB) are unpredictable garbage. Can't really support those beyond, "get a better one." Expect to spend about $100 per channel for name-brand USB, or slightly less per channel for a multi-input PCIe card.

Display and Window capture are free, if your machine can run both OBS and the thing(s) that it's capturing.

 

[u666sa]

Cheap HDMI captures (~$20 USB) are unpredictable garbage.

Ordered this one. It is based on MS2131, chip specs here. Either way you spin it, I need a capture card for my microscope camera. So I will test it with screen, and if I like it, I will get another one. Looking at this card, what's your verdict? I will get my hands on it only at the end of month.

 

[AaronD]

Ordered this one.

AliExpress is not promising, regardless of what the thing actually is. And their picture of the case looks almost exactly like the deceptive ones that I got from Amazon before I knew about the problems.

It is based on MS2131, chip specs here.

If those specs are actually what you get, maybe. But like I said, I wouldn't trust anything cheap, or from cheap sellers like Ali is.

I would not guarantee that that chip is in that device. It might even be counterfeit, made to look like and report itself as the real thing, but not meet the specs. Only a detailed test, off-board, and possibly a destructive teardown of both the suspect and a known-genuine chip, can tell you that for sure.

I had a brief stint at an independent company that tested chips, and one of their services was counterfeit detection. I wrote the software to run some of those tests. (poke/prod/stress the chip like <this>, and it must respond like <that>, within a fairly tight window...)

Looking at this card, what's your verdict?

Maybe you get something good by accident, maybe you don't. And by the time you replace enough to get the accidentally-good one, you could have easily spent enough to have bought a known-good one the first time, except you still don't have the known-good one.