• Please review our updated Terms and Rules here

*new* 8088 corr. video!!!!

I'm flattered, but have you seen the new player and new format over at http://www.oldskool.org/pc/8088_Corruption ? Source for the player is available so the new format should be easy for you to adapt to (just slight header change) and the new player is 33% faster. I also include a very I/O intensive player that is 100% faster, but only if your CPU is significantly faster than your hard disk.

I mention this because your video data is in the old format. Wasn't sure if you saw the new videos on the site yet :)
 
Trixter said:
I'm flattered, but have you seen the new player and new format over at http://www.oldskool.org/pc/8088_Corruption ? Source for the player is available so the new format should be easy for you to adapt to (just slight header change) and the new player is 33% faster. I also include a very I/O intensive player that is 100% faster, but only if your CPU is significantly faster than your hard disk.

I mention this because your video data is in the old format. Wasn't sure if you saw the new videos on the site yet :)
Click to expand...

no i haven't actually, i'm going to take a look. btw you got 8088roll'd. :)
 
i have been working on an improved encoder a little bit already, although it's not that much improved but at least it isn't just 5 shade characters. can you explain a little more about how your current encoder does the job? i know it gets complicated.

what i was thinking of trying is actually having a table containing each ASCII character's 8x8 pixel map. then it'd take a 320x200 input, and compare each 8x8 block pixel by pixel and attempt to determine the closest match from the ASS-KEY table. my initial thought is XOR each corresponding pixel, adding the result of each operation to a value and determining the lowest value of the 256 sums. this of course would determine which character to use.

i have the feeling yours is much more complex than this. (and this wouldn't do color yet.. although it could be modified a bit for it)
 
trixter, i re-wrote the encoder to actually compare each 8x8 block from a 320x200 video with each 8x8 ASCII block and find the closest match... it's still not color but i am going to work on adding that. tell me if you think this looks any better:

(it only chooses from a certain range of ASCII chars, not all of them. i thought it looked weird that way.)

http://rubbermallet.org/8088roll.zip



EDIT: when i add color i'm going to go ahead and add your header type to the file.
 
Last edited:
Mike Chambers said:
can you explain a little more about how your current encoder does the job?
Click to expand...

Why yes, I certainly can!

what i was thinking of trying is actually having a table containing each ASCII character's 8x8 pixel map. then it'd take a 320x200 input, and compare each 8x8 block pixel by pixel and attempt to determine the closest match from the ASS-KEY table. my initial thought is XOR each corresponding pixel, adding the result of each operation to a value and determining the lowest value of the 256 sums. this of course would determine which character to use.
Click to expand...

That was my initial choice too. Watch my talk to see how it worked.
 
Mike Chambers said:
tell me if you think this looks any better:
Click to expand...

It looks better. If you're going to concentrate on B&W, consider using the darkgray and lightgray colors to give you more shades.

One dirty little secret I didn't mention in the talk is that all of the videos I have provided in "8088flex" format have been processed by me to remove noise, zoom in on subjects so they fill the frame, adjust levels, etc. You may want to do something similar in your source frames before encoding them, maybe in your encoder itself so you don't have to fiddle with virtualdub, premiere pro, etc.
 
man, i've been working on color. it's working. the tron disc scene is about halfway encoded as i type this. it's slow.

tell you what, i have a WHOLE new appreciation for your encoding techniques now after trying to make my own. so far it's looking roughly half as good as yours. you can still tell whats going on, and the colors are right but yours seems to use them more than mine. like the ground stays black instead of a dark blue and stuff like that. you'll see when it's done, i think it's still coming out pretty good.

i'm not working with background colors yet, that is probably why. also, out of curiously which ASCII values does your encoder select from? mines does:

16 thru 19, 32, 174 thru 178, 219 thru 223, and 254.

EDIT: download time

here you go let me know what you think please!

http://rubbermallet.org/miketron.zip
 
Last edited:
if anybody is too lazy to download 8088flex and play the file, :p i encoded it playing in DOSBOX to an FLV file which doesn't look as good but you can watch it from your browser:

http://rubbermallet.org/miketron.html


btw i calculate that you can fit about 2.8 hours of 8088flex footage at 24 FPS/22khz/8-bit/mono on a 700 MB CD-R... maybe we could start a business selling full movies on CD for vintage computer fans to watch on 27 year old machines. bwahaha. (just kidding)

i can see it now "was that an airplane?" "no i think it was a cat"





anyway trix, even though it's not top quality quite like yours if you'd like i can work on making this encoder read right from AVI files (it uses seperate image and raw audio files now) and spice up the GUI a bit and you could supply it for people to download if you want. or release yours i guess! my "GUI" is ugly looking ATM, but i will work on it. it's also unfortunately done in VB6. it takes 1 second to encode each frame, that's on my 3.4 GHz pentium 4 w/ HT.
 
Last edited:
Mike Chambers said:
tell you what, i have a WHOLE new appreciation for your encoding techniques now after trying to make my own. so far it's looking roughly half as good as yours. you can still tell whats going on, and the colors are right but yours seems to use them more than mine. like the ground stays black instead of a dark blue and stuff like that. you'll see when it's done, i think it's still coming out pretty good.
Click to expand...

It looks good, you're definitely improving!


i'm not working with background colors yet, that is probably why. also, out of curiously which ASCII values does your encoder select from? mines does:
16 thru 19, 32, 174 thru 178, 219 thru 223, and 254.
Click to expand...

I use all 16 colors (the 8088flex player automatically turns off the blink attribute so you can use all 16 too) and I compare with *all* 256 ASCII characters. Don't arbitrarily limit yourself just because you think some characters look "weird". Even with the best converter, 40x25 textmode *still* frames look terrible but when they're in *motion* they're fine.

Correction: I don't compare against all 256 chars; some are omitted to speed up the encoder slightly. For example, there's no reason to compare #0, #32 and #255 because they're all null font data. I also don't compare redundant color combinations; for example, #177 the shaded character, there's no reason to compare fg1 bg2 and bg1 fg2 because it is the "same". But those are optimizations in the encoder for speed, not quality.

And my encoder doesn't actually compare individual characters (anymore). You'll remember from the talk video that what it actually compares are 4x4 resampled characters against 4x4 resampled source picture blocks, which I found to be the best compromise between picture detail and color quality.
 
Mike Chambers said:
anyway trix, even though it's not top quality quite like yours if you'd like i can work on making this encoder read right from AVI files (it uses seperate image and raw audio files now) and spice up the GUI a bit and you could supply it for people to download if you want. or release yours i guess! my "GUI" is ugly looking ATM, but i will work on it. it's also unfortunately done in VB6. it takes 1 second to encode each frame, that's on my 3.4 GHz pentium 4 w/ HT.
Click to expand...

Ours are not that different. Mine accepts raw .TGA files; I convert the .WAV to 8-bit PCM using Audition or whatever other sound editor I have available, massaging it along the way (dynamic compression, noise removal, etc.). I then use a small program to assemble the converter output and pcm into a .tmv. And mine takes 2 seconds per frame on a 2.6GHz P4.
 
Trixter said:
Ours are not that different. Mine accepts raw .TGA files; I convert the .WAV to 8-bit PCM using Audition or whatever other sound editor I have available, massaging it along the way (dynamic compression, noise removal, etc.). I then use a small program to assemble the converter output and pcm into a .tmv. And mine takes 2 seconds per frame on a 2.6GHz P4.
Click to expand...

same here. 8-bit sound needs all the help it can get. i use cool edit pro 2.1, and do dynamic compression and amplify before i turn it into 8-bit mono.
 
here's a shot of my encoder in action:

encoding.png




aaand the original frame from that particular ASCII conversion:

new1046.jpeg



(the 159 left means minutes)

it jumped way up to 5.7 or so seconds per frame rather than just over 1 when i told it to start looking at all ASCII chars instead of a smaller selection of them. ouch. in that pic, it's encoding an old live genesis video from 1980.
 
i tweaked the encoding a bit... here's a new shot. i'll have this video up later tonight. it's coming out realllly good so far:

encode8088.png


and it's a good song. squonk performed live in 1980, if anybody likes this type of stuff.
 
Trixter said:
Here's mine, for reference:
my_example.jpg


BTW make sure you've got the right colors for color index #6. It's brown, not dark yellow. See the wikipedia entry for CGA for values.
Click to expand...

i used the wikipedia values actually. i think just windows shows it weird for some reason. damn yours looks much better.... i should just quit now hah. i think what i need to do is have it identify the second strongest color, and make that the background attribute like attributebyte = (bgcolor * 16) + fgcolor ????

that oughta help tremendously.
 
ok i think i've done some massive improvements on detail and color.. check this out. oh and i might release my encoder to the public soon after much more GUI work and possible AVI support:

newencoder.png



what do you think??

the color definition still isn't as good as yours at all trixter, but i think i'm going to watch your explanation video again for pointers. i think think it comes out watchable from my program.
 
Last edited:
if i were to make a thread where my encoder could be downloaded, should i do it in this forum or vintage software/programming? i'm not sure where it fits best.. the actual encoder is obviously for win32 even though it's output is for vintage 8088's.
 
/me postwhores more

here is the tron disc scene encoded with the latest compile of 8088enc, i think muchhhh better. maybe it's just me and wishful thinking, but the colors seem much more accurate and vibrant, and the floors and walls arent all black anymore.

http://rubbermallet.org/newtron.zip
 
I'm at work so it will be six hours before I can check this, but it tickles me that you're working on it :)

Saw your encoder post in another thread; would you like me to alter the website to include it, or are you still working on it?
 
You must log in or register to reply here.
Back
Top