VOGONS


VGA Capture Thread

Topic actions

Reply 562 of 1396, by xjas

User metadata
Rank l33t
Rank
l33t

Here's a preliminary shot at Dreamcast VGA capture. I have to play with this more to get the settings right... for some reason all the games were being slightly undersampled resulting in pillarbox bars. Unfortunately I didn't realize this was happening until after the fact, which is why the screenshots are 597x480. On the other hand, the menus and splash screens filled the frame but the whites came in way dark. (Also, my cheapest-possible-one-off-Ebay VGA cable is a piece of junk and might be causing color / intensity issues due to bad contacts. Ordered another one from a slightly more reputable vendor; will report back when that gets here.)

Anyway, not bad for a first attempt:

CameraZOOM-20181105032024491.jpg
Filename
CameraZOOM-20181105032024491.jpg
File size
180.39 KiB
Views
1391 views
File license
Fair use/fair dealing exception
jgr1.jpg
Filename
jgr1.jpg
File size
388.04 KiB
Views
1391 views
File license
Fair use/fair dealing exception
daytona3.png
Filename
daytona3.png
File size
521.74 KiB
Views
1391 views
File license
Fair use/fair dealing exception
cracktro.png
Filename
cracktro.png
File size
392.21 KiB
Views
1391 views
File license
Fair use/fair dealing exception

^^ shockingly, not all my Dreamcast games are legit.

zerogunner1.png
Filename
zerogunner1.png
File size
450.68 KiB
Views
1391 views
File license
Fair use/fair dealing exception

Full gallery (lossless, only cropped)

Incidentally, seems like the VisionRGB Pro PCI can't quite sustain 30FPS at 640x480; I get the odd dropped frame. Forget about 60FPS. My recommendation is if you're going to game directly through the capture window or want "perfect" 60FPS captures, go for the PCI-express cards. I'm not using world's most powerful PC to capture though.

twitch.tv/oldskooljay - playing the obscure, forgotten & weird - most Tuesdays & Thursdays @ 6:30 PM PDT. Bonus streams elsewhen!

Reply 563 of 1396, by xjas

User metadata
Rank l33t
Rank
l33t

A few more for those who don't feel like clicking through to the gallery.

zerogunner2.png
Filename
zerogunner2.png
File size
546.33 KiB
Views
1391 views
File license
Fair use/fair dealing exception
armada2.png
Filename
armada2.png
File size
631.11 KiB
Views
1391 views
File license
Fair use/fair dealing exception
timestalkers3.png
Filename
timestalkers3.png
File size
424.14 KiB
Views
1391 views
File license
Fair use/fair dealing exception
ut99.png
Filename
ut99.png
File size
452.85 KiB
Views
1391 views
File license
Fair use/fair dealing exception
rally3.png
Filename
rally3.png
File size
512.78 KiB
Views
1391 views
File license
Fair use/fair dealing exception

BTW VCF1.2 is working great, some really nice features compared to Datapath's stock util. (And it's totally possible to 'tune out' the sampling issues I had, I just hadn't done it when I took all these shots.) Huge props to vvbee. 😀

twitch.tv/oldskooljay - playing the obscure, forgotten & weird - most Tuesdays & Thursdays @ 6:30 PM PDT. Bonus streams elsewhen!

Reply 564 of 1396, by leileilol

User metadata
Rank l33t++
Rank
l33t++

😁 Thanks for the VGA console PowerVR2 porn! Looking now.

- ut99-2.png's a great shot. It shows off the stepped bilinear filtering on the lightmaps and the close wall texture and looks similar to PCX2 and Kyro, as suspected.
- timestalkers2.png shows that off as well
-Familiar post-dither everywhere matches PCX2 and Kyro 😀 timestalkers1.png and cracktro.png look beautiful even if the downsampling seems to eat them.
-the Jet Grind Radiooooooooo shots show off the mip LOD

Probably the only request I have in mind are more alpha blend edge fade closeups. The only example that immediately comes to mind is Tails crashing in SA1 since there's a camera panning close to the plane that's dropping a bunch of alpha blend (to the point the fillrate chokes). Firing off rockets and grenades, and plasma/rail marks fading out in Q3 would show it off as well as the skies (even better, that game has a bindable zoom view action)

apsosig.png
long live PCem

Reply 565 of 1396, by vvbee

User metadata
Rank Oldbie
Rank
Oldbie
xjas wrote:

Incidentally, seems like the VisionRGB Pro PCI can't quite sustain 30FPS at 640x480; I get the odd dropped frame. Forget about 60FPS. My recommendation is if you're going to game directly through the capture window or want "perfect" 60FPS captures, go for the PCI-express cards. I'm not using world's most powerful PC to capture though.

On my system, it does solid 60 fps at 640 x 480 and 45 at 800 x 600 with 32-bit color, and that's running through a virtual machine.

Some months back I implemented run-time memory validation in vcs, which can have performance implications. It didn't on my test system, but beyond that I rely on reports from users. Vcs should in any case be slightly faster than datapath's program, so if that's noticeably not the case, let me know (having made sure you have the same color depth etc. when comparing).

Reply 566 of 1396, by xjas

User metadata
Rank l33t
Rank
l33t
vvbee wrote:

On my system, it does solid 60 fps at 640 x 480 and 45 at 800 x 600 with 32-bit color, and that's running through a virtual machine.

Some months back I implemented run-time memory validation in vcs, which can have performance implications. It didn't on my test system, but beyond that I rely on reports from users. Vcs should in any case be slightly faster than datapath's program, so if that's noticeably not the case, let me know (having made sure you have the same color depth etc. when comparing).

Hmm, it's hard to directly compare as I can't find any way to signal dropped frames in Datapath's util, but I'm definitely getting a bit better performance there than in VCS. I have the colour sampling set to 8-8-8 in both and am not scaling the image. I'd estimate at 640x480x60 VCS slows down or drops frames about 25-30% of the time, Datapath's util ~10%.

(I can do a 60FPS screen recording of both running and post it if that's useful.)

What's your test setup? I'm running this on an Athlon 64x2 4800+ / 4GB PC2-6400 / GeForce 8500GT. Not a particularly fast system, but should do the job?

I'm also getting a "stuck image" problem when I start VCS with the video source already active:

vcs-error.jpg
Filename
vcs-error.jpg
File size
549.44 KiB
Views
1349 views
File license
Fair use/fair dealing exception
vcs-error-2.jpg
Filename
vcs-error-2.jpg
File size
357.34 KiB
Views
1349 views
File license
Fair use/fair dealing exception

It will either get a single frame of video and stick there, or screen garble. Only power-cycling the Dreamcast (i.e. stopping and starting the VGA signal) will fix this.

I haven't tried to duplicate this behavior using other VGA sources yet.

twitch.tv/oldskooljay - playing the obscure, forgotten & weird - most Tuesdays & Thursdays @ 6:30 PM PDT. Bonus streams elsewhen!

Reply 567 of 1396, by Neco

User metadata
Rank Member
Rank
Member

a little hesitant to pick up a DGC103C because it seems like it might be a little underpowered. I want something I can grab a full 60fps with in DOS/Windows at higher resolutions.
$25 and local too so shipping would be fast, but still... I got enough frivolous toys for now I guess.

This Gefen would be a lot better if it took multiple input sources (curious to see how it would scale other game console output)

I'm assuming I want to look at something like an RGB E1S ?

Reply 568 of 1396, by vvbee

User metadata
Rank Oldbie
Rank
Oldbie
xjas wrote:

What's your test setup? I'm running this on an Athlon 64x2 4800+ / 4GB PC2-6400 / GeForce 8500GT. Not a particularly fast system, but should do the job?.

That's a little too slow I'd say. I run vcs in a virtual machine with limited gpu, so I've had it rely more on the cpu. I use it with a xeon e3-1230 v3, which is a fair bit faster. I'm surprised a 4800+ runs it even that well - I remember mine struggling to render websites a few years back. Since vcs is open source, maybe someone implements streaming frames onto an opengl surface for what might be better performance in cpu-limited situations. Qt's opengl won't even launch for me in the vm, so.

I've seen the stuck frame issue, or one kind anyway, with the capture card in general. From what I remember, just toggling between another input resolution and back in vcs should unstick it. Don't know what causes it.

Reply 569 of 1396, by xjas

User metadata
Rank l33t
Rank
l33t

^^ cool, I'll try something faster (C2Q Q9300, pretty much the most powerful thing I have with expansion slots) and report back. I was intending to move the capture setup elsewhere and sell the Athlon machine anyway.

twitch.tv/oldskooljay - playing the obscure, forgotten & weird - most Tuesdays & Thursdays @ 6:30 PM PDT. Bonus streams elsewhen!

Reply 570 of 1396, by vvbee

User metadata
Rank Oldbie
Rank
Oldbie

I put in some basic opengl support. To try it, get that in the vcs directory, [temporary link removed; if you missed it, grab the vcs code off github and compile at your leisure with the USE_OPENGL define], and see what happens when you run it. I've some doubts about it being useful, since I can't test it properly with capture in the vm. It's slower than the cpu version in linux without capture, about 500 fps vs 1400 at 640 x 480. Overlays won't work with this version.

My advice is to have a beefy cpu, though, as vcs was made to offload to it as much as possible.

Last edited by vvbee on 2018-11-08, 16:58. Edited 1 time in total.

Reply 571 of 1396, by appiah4

User metadata
Rank l33t++
Rank
l33t++

Has anyone so far tried either of these cheapo converters?

vga-to-hdmi-cevirici-donusturucu-converter-adaptor__0445066290390899.jpg

vga-to-hdmi-cevici-donusturucu-sesgoruntu-100-uyum-garantili__1425895104447141.jpg

I'm aware that they will MOST LIKELY be crap but sometimes among those cheap stuff something comes along that performs more than adequate.. For example, I use this for SCART to HDMI for my AMIGAs and it works fairly well, some motion interpolation artifacts and 2 frames of lag but nothing I can't live with.

41bsO9MXl4L.jpg

Last edited by appiah4 on 2018-11-08, 20:31. Edited 1 time in total.

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 572 of 1396, by ruthan

User metadata
Rank Oldbie
Rank
Oldbie

appiah4 - im curious about these too, its a shame that we have zillion Youtubers which are producing almost same content, but nobody afaik is not doing this stuff, maybe because is just OEM, no advertising, probably no free samples, support etc.. and you need to have some video processing knowledge.. I quite similar with home networking stuff like switches / routers, there is very hard to find proper review, because if someone review that its usually some highend "gaming" models.

Im old goal oriented goatman, i care about facts and freedom, not about egos+prejudices. Hoarding=sickness. If you want respect, gain it by your behavior. I hate stupid SW limits, SW=virtual world, everything should be possible if you have enough raw HW.

Reply 573 of 1396, by ruthan

User metadata
Rank Oldbie
Rank
Oldbie
vvbee wrote:

If you just read the whole thread, you could summarize the contents so far. Pros and cons people mentioned, etc. Put it on the wiki or wherever and get it going.

Here you go, i created this sheet primarily for myself, because its really too much information here and its hard to keep all of this in memory for longer time and tried to transfer information from this thread and sources sources into it:
link // It has lots of tabs for difference type of devices.

I tried to map info best as possible, but some information are in different formats for different manufacturers and i dont really understand technical background behind it (there are lots of cells with comments with things to solve), so someone more fluent in capturing can fix it, just write new PM i will give you writing rights. I just created some basic list and framework, now is experts time..

I also tested quite a lot with composite / s-video outputs in DOS, to check which resolutions are working and which ports are working when are these active.

Im old goal oriented goatman, i care about facts and freedom, not about egos+prejudices. Hoarding=sickness. If you want respect, gain it by your behavior. I hate stupid SW limits, SW=virtual world, everything should be possible if you have enough raw HW.

Reply 574 of 1396, by xjas

User metadata
Rank l33t
Rank
l33t
appiah4 wrote:
Has anyone so far tried either of these cheapo converters? […]
Show full quote

Has anyone so far tried either of these cheapo converters?

vga-to-hdmi-cevirici-donusturucu-converter-adaptor__0445066290390899.jpg

A while ago I bought the three cheapest VGA-to-HDMI converters I could get off Ebay, including the white one, intending to do a review/round-up. To be honest, there isn't that much performance difference between them and at least two of the ones I got are based on the same chip.

The thing you need to understand about these is they are not upscalers; the HDMI signal they create is the same resolution as the VGA signal they take in. Your VGA output already needs to be in an HDMI-compatible mode, i.e. 640x480 @ 60Hz, 800x600 @ 60Hz, 1280x720, etc. VGA text mode (400p @ 70Hz) is problematic as that's not an HDMI standard, but I haven't confirmed it universally doesn't work either.

I still haven't finished the writeup; I was going to test various 70Hz+ modes and tweakmodes and see what happens. I'll put some work into it when I get more time.

vvbee wrote:

I put in some basic opengl support. To try it, get that in the vcs directory, [temporary link removed; if you missed it, grab the vcs code off github and compile at your leisure with the USE_OPENGL define], and see what happens when you run it. I've some doubts about it being useful, since I can't test it properly with capture in the vm. It's slower than the cpu version in linux without capture, about 500 fps vs 1400 at 640 x 480. Overlays won't work with this version.

My advice is to have a beefy cpu, though, as vcs was made to offload to it as much as possible.

Thanks for this! I'll give it a try shortly. Is there any way to "benchmark" VCS, i.e. see a simple percentage of how many frames are dropped or how much time it spends dropping frames vs. normal performance? Would be useful for tweaking my capture machine. I have a Phenom II on the way and some much faster RAM left over from my other AM2 project.

twitch.tv/oldskooljay - playing the obscure, forgotten & weird - most Tuesdays & Thursdays @ 6:30 PM PDT. Bonus streams elsewhen!

Reply 575 of 1396, by vvbee

User metadata
Rank Oldbie
Rank
Oldbie
xjas wrote:

Thanks for this! I'll give it a try shortly. Is there any way to "benchmark" VCS, i.e. see a simple percentage of how many frames are dropped or how much time it spends dropping frames vs. normal performance? Would be useful for tweaking my capture machine. I have a Phenom II on the way and some much faster RAM left over from my other AM2 project.

If the refresh rate is 60 hz and you're getting 30 fps, you're dropping 50%. But basically you want the frame rate to be the same as the refresh rate.

I made a windows tool a while back to find the visionrgb's raw output fps, http://tarpeeksihyvaesoft.com/soft/?d=vrgb_fps.1_0:w. It does nothing but sit on the command line and count the number of frames per second the card is sending, so this gives you a baseline of how fast it can capture and deliver, without any extra processing for scaling or display. To test the first input channel with 32-bit color capture, run the program on the command line with the parameters 0 32. It'll keep printing out the fps about once per second until you ctrl+c out of it. Set the capture source to whatever resolution you want, first, and make sure you don't have other captures (e.g. vcs) running simultaneously, or you'll be slashing the output fps.

ruthan wrote:

Here you go, i created this sheet primarily for myself, because its really too much information here and its hard to keep all of this in memory for longer time and tried to transfer information from this thread and sources sources into it:link // It has lots of tabs for difference type of devices

That's a good start there. Some people, like the retro demoscene folks, want to know about nonstandard resolutions and refresh rates, so maybe some info on those would be useful. The visionrgb-pro can do custom modes based on my brief testing, and probably the e series too. But apparently some brands can't. Then there's the question of how to best integrate sample videos and screenshots into the text. I guess cherry-picked links can work, too. Maybe agree on some standard thing to capture for a demonstration.

Reply 576 of 1396, by ruthan

User metadata
Rank Oldbie
Rank
Oldbie
xjas wrote:

The thing you need to understand about these is they are not upscalers;

Yeah, but there still pretty cheap boxes which claims to be upscalers like this:
https://www.ebay.com/itm/Active-HDMI-to-VGA-S … vIFe:rk:22:pf:0
https://www.ebay.com/itm/VGA-RL-Audio-to-HDMI … JMRQ:rk:18:pf:0
https://www.amazon.co.uk/dp/B00ATNF346
Because other all other here recommended stuff were still 100$+

vvbee wrote:

Then there's the question of how to best integrate sample videos and screenshots into the text. I guess cherry-picked links can work, too. Maybe agree on some standard thing to capture for a demonstration.

Pictures in cells suck, im not sure if Google sheet can showlike Excel floating picture on moveOver event, other would be better just link some online gallery. Videos are fine, link is enough, for me personally Youtube compression in now days is good enough so i dont need some raw no YT samplers, but someone could have different opinion.. best solution would be probably if we can test same videos on different cards and it on YT and in description link not YT footage in better quality - i think that it would be Win-Win solution - we "only" need someone to produce videos and tell which footage is the best to show quality difference - i theory we can create some testing packages with free demos / games demos and batch file for execution.. to make recording more simply and every footage the same..

Im old goal oriented goatman, i care about facts and freedom, not about egos+prejudices. Hoarding=sickness. If you want respect, gain it by your behavior. I hate stupid SW limits, SW=virtual world, everything should be possible if you have enough raw HW.

Reply 577 of 1396, by vvbee

User metadata
Rank Oldbie
Rank
Oldbie

I find getting 640 x 480 source onto youtube in good quality is non-trivial. You have to introduce a degree of blur or the pixels scale like garbage, but then you have a degree of blur which you don't want. You certainly lose a bunch of quality there. Low resolutions are notably easier, as are high ones.

Reply 578 of 1396, by ruthan

User metadata
Rank Oldbie
Rank
Oldbie

I have some heretical question.
What is so complicated on capture of digital signal, that these boxes are quite expensive? In Simple view of problem, you just need some device, which will make screenshots -for digital signal just dump some memory blocks - every 16 ms for 60 FPS and just send it to USB interface and all other processing if needed could be done by software by computer CPU.

Im old goal oriented goatman, i care about facts and freedom, not about egos+prejudices. Hoarding=sickness. If you want respect, gain it by your behavior. I hate stupid SW limits, SW=virtual world, everything should be possible if you have enough raw HW.

Reply 579 of 1396, by vvbee

User metadata
Rank Oldbie
Rank
Oldbie

I thought solving the retro vga capture problem was pretty easy and cheap. I just read through the thread, picked up the visionrgb-pro2, and was good. Writing proper software for it took a bit of time, as did setting it up for capture in linux that it had no support for. But the hardware itself was a piece of cake to round up.