Old dos demo(s) that still need a winport or videocaptures.
category: general [glöplog]
enigma: One suggestion: When you upload to youtube, put the source system (486slc, 386-40, etc.) that you captured from. It will help to evaluate the quality of the demo for those who can't run it on hardware.
Added...
this one please :'3
If you suggest demos for capture, please give also information on which hardware this should run. This saves me a lot of time, especially since some productions don't mention any requirements at all. Then I have to spend my time to look up the party, find out compo machine specs etc...
enigma: thanks for the captures. When you post links to youtube videos on pouet, can you submit them as an external link (click on "click here" at the bottom of the demo page) instead of adding a comment? It's easier to find them that way, instead of scrolling through the comments.
Mystic by TSI - reboots in both DOSBox and PCem. From the info file:
System requirments: i80386 CPU (or something better)
VGA card (tested on Paradise, Trident, Cirrus)
ca. 610K free mem
We recommend: a SoundBlaster (on 220h and IRQ7)
Some bug may appear 'coz there is no two similar PC on the world so we couldn't
test all configuration only the next things: 386DX/40Mhz, 486DX/33Mhz, some SB
clones, and the VGA cards what we mentioned.
System requirments: i80386 CPU (or something better)
VGA card (tested on Paradise, Trident, Cirrus)
ca. 610K free mem
We recommend: a SoundBlaster (on 220h and IRQ7)
Some bug may appear 'coz there is no two similar PC on the world so we couldn't
test all configuration only the next things: 386DX/40Mhz, 486DX/33Mhz, some SB
clones, and the VGA cards what we mentioned.
Ok, also added to most of the last captures a link to the video in high quality (e.g. full 70 fps, no youtube degradation) in the youtube description. If you don't like to click through everything, you can grab them also directly from ftp://78.46.141.148/videos/. VLC is able to stream and seek directly from an ftp source.
@enigma^2: THANK YOU A LOT \:D/
omg, i havn't seen our intro since we released it something like 14 years ago :')
btw, i didn't have sound on the video :(
it's my computer setup who suck at playing the video? (others video have sound, I tested with c64 krestage video).
if I remember well, the intro played sound only with a GUS...
omg, i havn't seen our intro since we released it something like 14 years ago :')
btw, i didn't have sound on the video :(
it's my computer setup who suck at playing the video? (others video have sound, I tested with c64 krestage video).
if I remember well, the intro played sound only with a GUS...
The video has a standard format with an audio stream:
Video: MPEG4 Video (H264) 640x480 59.7fps
Audio: AAC 96000Hz stereo
There is no Krestage video on the server. However if you speak of the Krestology video, then it has the same format.
Video: MPEG4 Video (H264) 640x480 59.7fps
Audio: AAC 96000Hz stereo
There is no Krestage video on the server. However if you speak of the Krestology video, then it has the same format.
ha yes you're right :)
so ok, it's probably a player problem on my computer :)
well, another one I REALLY want to see again!
public demand by Purple from 1999, dos/gus intro, tech requirement in the nfo are really low, so i think it will be ok :)
can't wait to see it again :D
so ok, it's probably a player problem on my computer :)
well, another one I REALLY want to see again!
public demand by Purple from 1999, dos/gus intro, tech requirement in the nfo are really low, so i think it will be ok :)
can't wait to see it again :D
I added a couple dosbox captures:
Mighty by Xtacy
Just by Legend Design - with and without blur on the fakemode parts.
I was also gonna add Lunatic by Extreme, then found out Vicious already made YT video for all the Extreme and Velvet releases. I still wanna make a video for Techno Holic that speeds up the logo/writer parts and slows down the effect parts.
Mighty by Xtacy
Just by Legend Design - with and without blur on the fakemode parts.
I was also gonna add Lunatic by Extreme, then found out Vicious already made YT video for all the Extreme and Velvet releases. I still wanna make a video for Techno Holic that speeds up the logo/writer parts and slows down the effect parts.
no update? :'(
public demand works fine in DOSBox, you can't watch it there?
The point of the thread was to request (and provide) captures for DOS demos that do NOT work properly in emulators.
The point of the thread was to request (and provide) captures for DOS demos that do NOT work properly in emulators.
@trixter: ahem... I read in the topic:
and "public demand" doesn't have any capture yet :D
I'm waiting that since 15 years, since I never get a hand on a DOS PC with GUS since mine died :3
Quote:
Old dos demo(s) that still need a winport or videocaptures.
and "public demand" doesn't have any capture yet :D
I'm waiting that since 15 years, since I never get a hand on a DOS PC with GUS since mine died :3
It's easy to setup dosbox with a GUS/GUSMAX output? (i'm under linux)
If it isn't enabled yet, set gus=true in dosbox.conf and be happy.
Quote:
and "public demand" doesn't have any capture yet :D
I captured it from DOSBox: https://www.youtube.com/watch?v=Jd2vSI9ahMw. Enjoy!
HA NEAT! thank you a lot phoenix :D
Just a small suggestion, when uploading to youtube, you should always scale it 2-4x from the original size or even 6-8x (with nearest) on the video as it will get rid of chroma subsampling artifacts and allows higher a bit rate with zero artificial smoothening/etc..
This is something I always do with any lowres stuff I upload.
https://www.youtube.com/watch?v=3vHIXWQTFA8 Compare this to "no upscaling" https://www.youtube.com/watch?v=C6vjnAJHJBk
This ain't a demo (keen1 mod instead) but it shows that even with the scaled one at "480p"quality setting it already suffers from blurring, especially around darker hues. Even though it's technically adding more pixels, no real filtering is ever being applied. At 720p quality it already looks near what you'd get from a very lightly lossy/compressed image
Video players can then use whatever resizing algorigthm to downscale it further as needed, just means that the footage arrives on the viewer's end in best possible quality. Scaling gets done regardless since nobody who watches youtube runs their desktop at 640x480 anyway.
I know it's a dirty workaround for sure :(
It's a bit of extra work but the quality difference when forcing higher bit rates can be really major and worth it in the end.
Plus it makes it more "future proof" as quality can still be considered high after a few years aside from the lack of 50/60fps.
I know that this can be rather easily scripted in virtualdub for example.
Generally a 4x upscale for 320x200 is what I would do at least. I think in the Keen video I upscaled it by 8x. Remember to always have it as an even multiplier due to chroma resolution being half of luma.
This is something I always do with any lowres stuff I upload.
https://www.youtube.com/watch?v=3vHIXWQTFA8 Compare this to "no upscaling" https://www.youtube.com/watch?v=C6vjnAJHJBk
This ain't a demo (keen1 mod instead) but it shows that even with the scaled one at "480p"quality setting it already suffers from blurring, especially around darker hues. Even though it's technically adding more pixels, no real filtering is ever being applied. At 720p quality it already looks near what you'd get from a very lightly lossy/compressed image
Video players can then use whatever resizing algorigthm to downscale it further as needed, just means that the footage arrives on the viewer's end in best possible quality. Scaling gets done regardless since nobody who watches youtube runs their desktop at 640x480 anyway.
I know it's a dirty workaround for sure :(
It's a bit of extra work but the quality difference when forcing higher bit rates can be really major and worth it in the end.
Plus it makes it more "future proof" as quality can still be considered high after a few years aside from the lack of 50/60fps.
I know that this can be rather easily scripted in virtualdub for example.
Generally a 4x upscale for 320x200 is what I would do at least. I think in the Keen video I upscaled it by 8x. Remember to always have it as an even multiplier due to chroma resolution being half of luma.
@Oasiz: I do this upscaling specifically for youtube by default, usually 2x.
As for the resolution and subsampling it is actually more complicated. Your example of 320x200 mode resulting in a 320x200 capture resolution is only true if the video comes out of an emulator like dosbox. Standard low res modes of a VGA card translate to 720x400. So on capture already subsampling on 8 bit RGB takes place as from an analog source the pixels are never ideally sharp. Now if you scale this by two you get a 1440x800 RGB image, where the option is to use mere pixel doubling or some form of interpolation as e.g. bilinear. As the source image is already subsampled the the argument for pixel doubling is not as strict any more.
The next issue is that the image is 4:3 which can be set in the container maintaining a multiple of the original resolution. This leaves the anisotropic output scaling to the player.
The color space conversion from RGB to YUV in encoding is not as critical in this aspect.
One more point is that originally on a CRT the pixel edges were also not ideally sharp. So for low res graphics a small amount of edge bleed, which can basically mapped to some interpolation on a higher resolution, may give even a better representation.
As for the resolution and subsampling it is actually more complicated. Your example of 320x200 mode resulting in a 320x200 capture resolution is only true if the video comes out of an emulator like dosbox. Standard low res modes of a VGA card translate to 720x400. So on capture already subsampling on 8 bit RGB takes place as from an analog source the pixels are never ideally sharp. Now if you scale this by two you get a 1440x800 RGB image, where the option is to use mere pixel doubling or some form of interpolation as e.g. bilinear. As the source image is already subsampled the the argument for pixel doubling is not as strict any more.
The next issue is that the image is 4:3 which can be set in the container maintaining a multiple of the original resolution. This leaves the anisotropic output scaling to the player.
The color space conversion from RGB to YUV in encoding is not as critical in this aspect.
One more point is that originally on a CRT the pixel edges were also not ideally sharp. So for low res graphics a small amount of edge bleed, which can basically mapped to some interpolation on a higher resolution, may give even a better representation.
Another thing I'd like to mention when capturing from real PCs is that the quality of analog filters on the RGB output can vary wildly from one VGA card to the next.
There is a low-pass filter on there, to make sure the card stays within legal emission/interference ratings etc. It mainly consists of a series of capacitors and resistors.
However, in order to keep costs down, many cards use very simplified circuitry. Combine that with low-grade components (too much tolerance), and the filters do not work as intended. Instead of just filtering out the frequencies outside the regular signal range, they also filter out high-frequency components of the signal itself. This means that high contrast between two adjacent pixels will result in blur. This gets worse as the resolution and/or refresh-rate increases, since this means the signal rate increases as well, effectively giving you higher frequencies in your signal.
Matrox had a strong reputation for having very high quality.
For other cards, a common hack back in the day was to remove part of the filter, so that the lower frequencies would no longer get filtered/blurred. I have modified an Asus GeForce2 GTS this way, back in the day, and it resulted in very crisp image quality, very similar to the Matrox cards I have.
I think it will be worthwhile to select cards based on their output quality (Matrox is always a good choice) and/or to modify their filters, because it will surely improve your capture quality. My own experiments with capturing revealed that my capture equipment was especially sensitive to the output quality. The Diamond SpeedStar PRO VLB I used in my 486DX2-80 gave a very noisy image. Other cards would give better quality using the exact same capture setup.
There is a low-pass filter on there, to make sure the card stays within legal emission/interference ratings etc. It mainly consists of a series of capacitors and resistors.
However, in order to keep costs down, many cards use very simplified circuitry. Combine that with low-grade components (too much tolerance), and the filters do not work as intended. Instead of just filtering out the frequencies outside the regular signal range, they also filter out high-frequency components of the signal itself. This means that high contrast between two adjacent pixels will result in blur. This gets worse as the resolution and/or refresh-rate increases, since this means the signal rate increases as well, effectively giving you higher frequencies in your signal.
Matrox had a strong reputation for having very high quality.
For other cards, a common hack back in the day was to remove part of the filter, so that the lower frequencies would no longer get filtered/blurred. I have modified an Asus GeForce2 GTS this way, back in the day, and it resulted in very crisp image quality, very similar to the Matrox cards I have.
I think it will be worthwhile to select cards based on their output quality (Matrox is always a good choice) and/or to modify their filters, because it will surely improve your capture quality. My own experiments with capturing revealed that my capture equipment was especially sensitive to the output quality. The Diamond SpeedStar PRO VLB I used in my 486DX2-80 gave a very noisy image. Other cards would give better quality using the exact same capture setup.
Video is hard!
Just captured Facts of Life by Witan. I think the linked youtube videos on the prod page are somewhat wrong, as they only show one part from an extra GUS version of this part.
The actual demo has a few more parts, starting from a star wars scroller and it even has loader parts. I used a 386DX-40, ET4000, SB16. It didn't mattered if I loaded QEMM or just HIMEM. I first tried it on a 486DX2-66 but the Vector part was running too fast.
Maybe someone who knows how the demo should look could view it once to check if everything ran well. (e.g. end scroller has no music?)
The actual demo has a few more parts, starting from a star wars scroller and it even has loader parts. I used a 386DX-40, ET4000, SB16. It didn't mattered if I loaded QEMM or just HIMEM. I first tried it on a 486DX2-66 but the Vector part was running too fast.
Maybe someone who knows how the demo should look could view it once to check if everything ran well. (e.g. end scroller has no music?)
Target system for Facts of Life was a 386-16. Do you have one of those? If not, let me know.
Well, I have a board with 386DX-25 that I can setup on a table. If I deactivate the L2 cache it will come close. I also have a 386-16 in a IBM PS/2 Model 80, but I couldn't find a working SB Pro MCA yet. And then there is a board with 386SX-16, though I haven't found the time to fix the Dallas Battery/RTC chip.