converting some usual AV file to STE hi-color HAV file
There are diverse STE HAV formats introduced by me - like
320x200, Overscan 416x228 etc. This guide gives general help for all
them, but as first example we will deal with conversion to format
12.5 fps , because it is most interesting at moment, for people
equipped with UltraSatan.
What we do is basically extraction of picture sequences from movies,
videos, then converting them with special SW into Atari STE hi-color
format. Need to extract audio too, and convert it to format usable on
STE. Final step is multiplexing hi-color pictures and audio into HAV
file - something like AVI for STE.
Our main tool will be Virtual Dub. It is very powerful, reliable,
freeware and has good support, so we have many plugins available - and
will need some. Note for Linux users: almost everything can be done in
Linux too, for sure. But I'm not familiar with Linux SW for this
purpose. Plus, SW for conversion into STE hi-color format is for
As first example, we will convert something simple and short. It is old
Mpeg1 video Test.mpg , suplied with Xing player, around 1995-6. Need
only Virtual Dub - it handles Mpeg1.
Open file Test.mpg and go to Video/Filters.
Select resize :
Then make following
We need 320x158px, but will add 1-1 black line to top and bottom,
because converters usually cut off top line. In some cases you will
need to Letterbox to 320x200 (depending on SW for conversion to
hi-color), then enter 320x200 in Framing
options . Then will have black areas at top and bottom.
OK.OK, and you will see something like:
Now we see
something, what is usual problem after
setting image size to output resolution (dictated by HW,
player res.): Aspect ratio error (output is at left here) .
In reality, it will be not so much stretched vertically on Atari
To correct AR errors we can use cropping of image edges
- in this case cropping at top and bottom, 16-16px for instance.
After it, we need to set proper frame rate:
In this case, it is
as is shown left. Playback will be slower some 20%,
because src. rate is 30fps, but we can not do better here. There are
which drop every 6-th frame to maintain same playback speed, but it is
ugly. And of course, change of playback speed will cause
desync - later about how to fix it. Here, it is not relevant, because
we have only
30 fps source is not common today - we will see mostly 24 fps. Or in
of TV recordings (Europe) 25fps. Why not keep 30 fps, and play on
with 30 or 15 fps ? It will not happen - NTSC video mode on Atari
very bad for hi-color video playback. There is no enough border space
Later, I will add some help how to do real and quality framerate
MVtools. You will need AviSynth too. It is not easy - command line, but
powerful. Mvtools generates new frames, based on motion analysis, what
much better than frame dropping/doubling combined with blending.
Next setting is important for good looking video on Atari STE:
We need to perform color reduction to 4096 colors, together with
error diffusion. There is special, very good Virtual Dub filter for it :
Set RGB bits to 4 as is shown. OK.
In some cases you don't need above - depending from settings
in conversion SW. Later more about ..
After setting all, we need to export image sequence :
Then, need to extract audio too:
Select 'Full processing mode' - so will save as PCM, what is certainly
compatible with audio editing SW.
We need yet to convert saved WAV to format suitable for STE. It is
25033 Hz samplerate, 8-bit mono, unsigned, as WAV file.
It can be done with diverse audio processing SW.
How to do it with freeware AudaCity, see here: STE background player section Preparing
WAV file on PC .
Slowest part is conversion of image sequences to STE hi-color
format, especially if want better quality. SW is developed by Douglas
Little - Photochrome, and Cyg - HigheSTcolor . Both is under
development currently, and signifficant improvement is made in last
couple months. You can DL latest test versions of HigheSTcolor
in forum of this site. And of course posting there your questions,
I will add here downloads in few days. Just be aware that all this is
very complex and hard, so don't expect miracles and fast results.
Next: usage of diverse plugins in Virtual Dub. You will need them
for mp4, mov, mkv etc. formats. AVIsynth, framerate
conversion with MVtools.
And: there are another ways. Ffmpeg is extremely powerful video/audio
processing SW. Can extract image sequences, audio, process them ...
Command line tool.
In same cases, I could extract audio only with Ffmpeg.
Old fashion Framerate conversion:
It is one of hardest parts in AV
processing, conversions. Need for framerate conversion appeared first
when started to show regular films on TV. In Europe, where TV standard
is 25 fps, showing films shot at 24 fps is solved pretty simple, and
without picture quality loss: speeding up playback speed to 25 fps. Of
course, then need to speed up audio too. All it was simple with
analogue technic. Bad side is 4% faster playback speed than speed of
recording, what means higher pitch of audio too - some people is
sensitive. But there was no other way practically.
In NTSC lands, solution was more complicated - speeding up from 24 to
30 fps would be too much, so they invented Telecine. If you duplicate
every 4th frame, will get 30 from 24 fps. But it results in jerky
movement. Therefore they used fact that TV frame is made from 2 fields,
which are interlaced, and duplicated only 1 field at once, what looks
much better. It is of course far from ideal, but that was possible in
Today, with modern TVs, monitors we can have diverse framerates, and
conversion is usually not necessary. What is in is upscaling fps at
playback, to 50-60 fps. Our old Atari STE is
capable for 25 fps, or 12.5 fps, and even 50 fps playback. 30 fps is
not really possible because smaller border area. So, in case of some
source with different fps, we need to convert. First example,
above. was not really good solution - playback speed is 20% lower. But
will see later how to do it better. Now, let see first most
common case: conversion from 24 fps to 25 fps. We will do same,
what TV stations did wuth films - speed up. Nothing special to do with
video - just extract image sequence normally. Extract audio with
Virtual Dub if possible. If it goes not, best way is to do it with
Ffmpeg : ffmpeg -i
src.avi out.wav (replace src.avi with your filename) - really not that hard,
We need to speed up audio, otherwise will loose sync. It can be
done with better audio processing/editing SW .
Proper audio conversion if src.
is 24 fps, with AudaCity :
In most of cases exact framerate of source is 23.976
fps - and better use exact value, so will get synced audio even by
Extract audio with Virtual Dub (or some other SW) - use full processing
mode for audio, as explained above. Open it with AudaCity.: Tracks/Stereo Track to Mono
by need (if stereo wav).
First thing to do is normalize. In some cases it is better to use
Compressor - if volume varies too much.
After normalisation it is little louder, so will be less noise.
In case of 23.976 fps source, we need Effect,
Change Tempo . Ratio is 25/23.976=1.042709, what makes
4.2709 % :
For 1500 frames we must get exactly 1 min duration audio. If doing 12.5
fps, then 1 min stays for 750 frames.
Next important thing is setting Project
Rate (Hz) to 25033 :
After it File,
Save as type: Other uncompressed files, click on Options , set WAV (Microsoft) ,
Unsigned 8 bit PCM . OK, Save
It is proper format for STE playback.
Advanced framerate conversion:
Quality framerate conversion without changing
playback speed, audio pitch, jerky movement (dropped/duplicated
frames), blending ? It is possible today.
There is freeware SW called MVTools, which can generate new frames,
based on motion analysis. Of course, this is not perfect - in some
cases it may look not good. DL AVISynth and MVTools2
and install them. Usage is not so
hard, as it may seem at first. Again, example video conversion - this
time Atari ST related. Best quality one, I found on WEB. It is 30 fps,
and we want to convert it to our 25 fps, or 12.5 fps (UltraSatan) ,
without slowing down playback.
AVISource("d:\acsPart.avi") # or MPEG2Source, DirectShowSource, some
previous filter, etc
super = MSuper(pel=1)
backward_vec = MAnalyse(super, overlap=4, isb = true, search=3)
# Use block overlap, halfpixel accuracy and Exhaustive search
forward_vec = MAnalyse(super, overlap=4, isb = false, search=3)
MFlowFps(super, backward_vec, forward_vec, num=25000, \
Only part you need to edit/change is in first line - replace path and
filename with your one. Save script with extension
Open it in Virtual Dub, and go to Video, Filters, add resize :
After setting resize need to crop - remove black parts left and right,
top, bottom. + because this will be 320x158 need to cut little
more at top and bottom.
Important to get correct AR - we don't want too much streched picture.
This is how output image will look. + File information - as you may
see, it is exactly 25 fps.
When all is set, we need only to extract image sequence. Only if want
12.5 fps playback need to go in framerate setting:
Same thing (12.5 fps) we could get by replacing num=25000
in script above.
Because generating new frames is complex, it will work little
slower, may take 5-10 min for 1 min video. I don't see any, even
smaller errors in this conversion.
And MVTools may be good for many other videos. People using it
for get 48-60 fps from 24-25 fps sources. I did Hobbit trailer 24 to 50
fps, and looks pretty well.
Where to get source for above ? DL
clip . You will need flv plugin for Virtual
Dub too. Best is to make intermediate file in AVI format before
framerate conversion, with noise filtering, cropping etc.
If you managed it, you just made better conversion than
usual ones, and my first conversion of this Atari video (20% slower
than org.) .
PP Dec. 2012 .
Send mail with
questions, critic etc. to site maintainer: Before it enter name
of animal on small picture: