Artex wrote on 2021-03-13, 00:22:So I've read this entire thread - wowza, lots of good info and much more complex than anticipated. I've bought (and returned) q […]
Show full quote
So I've read this entire thread - wowza, lots of good info and much more complex than anticipated. I've bought (and returned) quite a few items (thank you Amazon) as I didn't read this thread first before buying and honestly thought video capture would have come a long way by now. However, capturing DOS resolutions @ 70hz really throws a wrench into things apparently.
I started with an Elgato HD60 S+, which I promptly returned after realizing that it doesn't do anything more than 60Hz (as per the case with other USB capture devices). I also ordered what was supposed to be a new-in-box StarTech.com VGA2HDMIPRO (https://www.amazon.com/gp/product/B001LRKB78/ … 0?ie=UTF8&psc=1) but Amazon sent me the newer version which doesn't handle the 70hz / 700x400 DOS mode as this thread has taught me, so that's going back as well.
My goal is to capture DOS games but also 3DFX Glide titles in Windows 98SE (1024x768) using my "Time Machine" box that has a TNT2 Ultra + SLI Voodoo 2 cards. So now have the following hooked up currently:
- Kaico Edition OSSC 1.6 (waiting on my 2GB MicroSD card to flash it to the latest firmware).
- UGREEN VGA Splitter 1 in 2 Out
- Elgato Cam Link 4K
So VGA out from retro box -> UGREEN VGA Splitter -> VGA Input Kaico Edition OSSC 1.6 -> HDMI Out -> Elgato Cam Link 4K which is plugged into a USB 3.0 Port.
I ordered a bunch of other stuff, currently en-route:
- DataPath VisionRGB-E1S with full-size bracket
- CableCreation DVI to HDMI Adapter
- Atlona AT-HD500 PC/Laptop to HDMI Converter with Built-In Scaler
I've tried OBS and VirtualDUB2, both of which seem to require you to reconfigure/re-add the device each time the video resolution or refresh rate changes. Do any of these items remain synced when going from DOS into Windows or when rebooting? Is this the norm??
It sounds like my best path going forward is:
VGA out from my retro box -> OSSC (RGB3 in Pass-Through Mode) -> HDMI Out to HDMI-to-DVI Adapter -> DataPath VisionRGB-E1S with full-size bracket in my main capturing rig (using the homebrew VCS app).
Can someone check my understanding that both the Atlona AT-HD500 and Elgato Cam Link 4K require conversion to RGB to use the ffull colorspace post-capture? Is there any benefit in replacing/swapping the OSSC with the Atlona AT-HD500? It seems like the OSSC is just doing passthrough to convert VGA to HDMI?
I'm totally green when it comes to this colorspace stuff (and the OSSC) so any help from you techies out there would be appreciated!
Looks like you are getting equipped to have fun .
First, I suggest you read (or at least skim through) the following four Wikipedia articles .
https://en.wikipedia.org/wiki/RGB_color_model
https://en.wikipedia.org/wiki/YCbCr
https://en.wikipedia.org/wiki/Chroma_subsampling
https://en.wikipedia.org/wiki/Color_space
a) Changes in video resolution/refresh normally require capture device reconfiguration in Virtual . However, if I understand correctly, using VCS on an E1S would allow you to use automatically switchable presets for different resolutions while scaling in real time to a video stream at a chosen target resolution that you then capture . See ( Re: VGA Capture Thread) . I have not tried this yet
b) Using an OSSC and splitting its output for HDMI is my personal preference so far. I like it's configuration options and have not had great luck with either VGA splitting or VGA directly into an E1S . I also prefer to keep the analogue signal chain as short as possible . AFAIU, using an OSSC in this way does not prevent you from using VCS and it's preset system, but I could be wrong).
c) I don't know what color processing the AT-HD500 uses internally. IMHO, its only practical use is converting 70Hz VGA to 60Hz DVI/HDMI and the Extron RGB-DVI 300 and RGB-HDMI 300 units are much better than the AT-HD500 if you need scaling and conversion from 70Hz to 60Hz . EDIT: If you are going to be capturing using the E1S, and are OK with with capturing 70Hz, using the OSSC to digitize will give better results than the AT-HD500 or the even one of the Extron units . You could capture VGA directly with the Extron, but you will likely need to use a VGA splitter to feed the monitor you will be watching to use the computer you are capturing from. Also, If you are going to be using the OSSC anyway for displaying on your retro machine's monitor, you might as well use the OSSC to digitize and split its output to the monitor and the E1S, which will have the added advantage of adjusting parameters such as phase on only one device (OSSC) rather than two (OSSC and E1S).
d) VGA outputs analogue RGB which need to be digitized for capture. The digitization and capture can be done by the same device (VGA capture device) or by separate cascaded devices (like an OSSC to digitize OR a VGA-to-HDMI converter/scaler to digitize AND a DVI/HDMI capture device for the actual capture). Either way, analogue VGA RGB video is digitized into digital RGB and then, depending on the device used for digitization, is either left as is in digital RGB format or is converted to digital YCbCr and then most often subsampled. The OSSC, for example, passes digital RGB to its HDMI output (though you can force to it output YCbCr, but you typically would not want that). I have not tested what the Atlona and Extron units output in term of digital signal and, even if they output digital RGB through HDMI/DVI, it does not guarantee that they process video internal in that format . As for VGA capture devices and HDMI/ones, most of them will not only force a conversion of received analogue RGB or digital RGB into digital YCBCr, but will also subsample to a 4:2:2 or even a 4:2:0 format .
e) Format conversion between digital RGB and digital YCbCr is not a lossless and can cause visible artifacts, even when there is no subsampling being done . Such conversions should only be done when necessary and avoided if possible .
f) Subsampling of digital YCBCr is to be avoided as much as possible during the digitization/scaling/capture phase of the process . Subsampling is particularly unkind to low resolution computer-generated video (especially if the images has hard well-defined edges) that has not been upscaled or preferably line doubled and will result in visible artifacts if done prior to upscaling or line doubling . Subsampling does serve a useful, even essential, purpose in final video encoding/compression .
g) The Datapath E1S is one of the few (affordable) devices than capture analogue RGB or digital RGB without converting to digital YCBCr or subsampling .
h) The Cam Link 4K does convert everything it receives to digital YCbCr and subsamples to 4:2:2 (which is better than 4:2:0 , but still).
i) As alluded to previously, line multiplying (of which line doubling is an specific case) or integer scaling, which are conceptually the same (and literally so when done in the digital domain), when done BEFORE YCbCr conversion and subsampling will greatly diminish the quality degradation due to subsampling . This is due to the nature of what subsampling is . See this post and the 3-4 following ones for some examples : Re: VGA Capture Thread
j) Once conversion to digital YCBCr and subsampling have occurred, there is no way reverse the effects of the related image quality degradation . Explicit conversion to RGB is always possible but will likely offer no tangible benefits at this point (except maybe it is done in order to apply a filter/effect that can only be applied to RGB video)
k) I have never used a non-linear video editor, except Virtual, so I do not know in which color format they usually operate, but I presume YCbCr . If you start off with digital RGB capture footage, it is probably best to upscale it to a higher resolution before importing into your editing application to maximize image quality. Someone with more experience/knowledge with this will hopefully chime in .
l) As mentioned in point f), digital YCbCr format and subsampling are useful and are nearly always used in video compression formats meant for use in content distribution . I believe the Wikipedia article on subsampling explains at least some of the rationale behind this .
I hope that helps answer some or hopefully most of your questions . Feel free to ask more.
Disclaimer : This is how I understand things . I have learned/practiced this in a DIY fashion, so I apologize for any possible factual errors or misconceptions and welcome constructive criticism . EDIT : Please do not make purchasing decisions solely based on my opinions .
EDIT : Corrected typos .
EDIT2 : added points k) and following