60 fps video using SIDE 2

[a8isa1]

50 minutes ago, Wrathchild said:

This all seems strange, I've never had to pad a resulting image file?

Phaeron says the movie starts at sector 16 [counting from 0] here, so I've always padded the file with 16 sectors .

[Wrathchild]

But that is handled within the mux program!

fseek(fo, 16*512, SEEK_SET);

 

[Guest]

2 minutes ago, Wrathchild said:

But that is handled within the mux program!

fseek(fo, 16*512, SEEK_SET);

 

what i thought too. seems pointless running mux if all it does is join aud/vid together...

wonder if it's different for real hardware as opposed to Altirra?

- 'cos real hardware is (thus far) refusing to play ball!

Edited August 14, 2019 by Guest

[flashjazzcat]

It should be just the same on real hardware. Have you got this working in emulation yet?

[a8isa1]

8 minutes ago, Wrathchild said:

But that is handled within the mux program!

fseek(fo, 16*512, SEEK_SET);

 

Now I'm confused   ?

 

I've done a few dozen conversions (with padding).  They've all worked PAL and NTSC in Altirra.  NTSC ones work on my real hardware (but have correct colors only on my scan converter).

Edited August 14, 2019 by a8isa1

[Guest]

11 minutes ago, flashjazzcat said:

It should be just the same on real hardware. Have you got this working in emulation yet?

no. same issues.

which is why i've persisted (for last few hrs) with real hardware

[flashjazzcat]

I'd get it working in the emulator first. It'll save a significant amount of time.

[+Stephen]

45 minutes ago, a8isa1 said:

Now I'm confused   ?

 

I've done a few dozen conversions (with padding).  They've all worked PAL and NTSC in Altirra.  NTSC ones work on my real hardware (but have correct colors only on my scan converter).

Been a while since I converted any, but I also do not remember ever having to pad the resulting image file.  I just used DD to place the raw image on the CF.

[flashjazzcat]

Mount the IMG file in the emulator. The will eliminate any errors when creating the disk in HxD. I'm wondering if the IMG file is being pasted into a logical volume on the CF card rather than to the raw physical media. This would do a fine job of misaligning the video data.

Edited August 14, 2019 by flashjazzcat

[a8isa1]

1 hour ago, Wrathchild said:

But that is handled within the mux program!

fseek(fo, 16*512, SEEK_SET);

 

totally puzzling.  video frames + audio are 17 sectors.  my padding is an initial 16 sectors. One of us should have scrambled results yet we both have working videos.

[_The Doctor__]

0-16 is 17, with data being pulled aligned on the 17th sector. do you suspect timing allows for + or 1 one sector?

[a8isa1]

13 minutes ago, _The Doctor__ said:

0-16 is 17, with data being pulled aligned on the 17th sector. do you suspect timing allows for + or 1 one sector?

I'm an idiot.  Disregard everything I said in my past 4 posts (or so)

 

I just tested. I did not think dd allowed seek=0 (no offset). Wrong!  I thought dd started counting at 1 so I was using seek=17 to set my offset, believed to be 16.

 

I read too much into Phaeron's description.  Wrong!

 

So I've made two mistakes and accidentally realigned the video on the disk.  Do two wrongs make a right?  I don't think so!

 

 

 

[Guest]

in addition to my PM sent, Avery, Steve and Jon : 

I would just like to say thanks guy for all your help - and (Jon) for his spotting my stupidity.

I'd been writing out the image to logical rather physical.

 

there won't be repeat offences - i'm taking myself outside and having myself shot at dawn!

 

thanks guys - much appreciated.

 

now, can the volume levels be increased?

[Wrathchild]

25 minutes ago, brenski said:

I'd been writing out the image to logical rather physical.

 

I had asked on Tuesday evening about this "I'm assuming this is using a tool such as HDDRawCopy (i.e. raw copy to sectors not file copied to FAT?)" and was told "have tried different utils win32 disk manager and HxD" so should have twigged you were actually doing this wrong ?

 

For volume, when converting the original video to the A8 avi there are some audio options you can add in to boost the volume (Audio/Volume... menu) but do that with care not to distort. You may also be able to edit the outputted RAW audio file within Audacity and process it within that.

 

 

 

 

[Guest]

cheers for this ^ really helpful (and apols - forgot to thank you in post above).  

 

it's not your fault that you missed my error - i claim that award all for myself 

?

 

i'm going to test this again now and see how far the volume levels can go without distortion - my guess? (from my old 1980s A8 sampling days) - not very far!

 

 

 

Edited August 15, 2019 by Guest

[Guest]

@phaeron @flashjazzcat @Wrathchild @Stephen @a8isa1

 

well, the highest i've found I can stretch the volume (without noticeable distortion) is 208.9%.

 

couple of questions:

■ will boosting the volume prior to initial (external) rendering - ie before Virtualdub and then less volume increase in virtual dub, make things sound better and louder?

■ is there any way at all to produce better video? again, would it help if the initial rendering was at (say) 720x405, 640x360 or even lower res, rather than 1920x1080?

 

these are just experimental ramblings...pretty sure this is going to be a case of Atari is already doing more than it should be able to?

 

would also be interest to hear what res @ndary video on page 7 was rendered at initially (ie prior to virtualdub being involved)

Edited August 17, 2019 by Guest

[+Stephen]

Been probably 2 years since I messed with any of this.  I don't think we'll be able to get much better visuals - maybe if there is some software for boosting the saturation before encoding?

[a8isa1]

11 hours ago, brenski said:

@phaeron @flashjazzcat @Wrathchild @Stephen @a8isa1

 

well, the highest i've found I can stretch the volume (without noticeable distortion) is 208.9%.

 

couple of questions:

■ will boosting the volume prior to initial (external) rendering - ie before Virtualdub and then less volume increase in virtual dub, make things sound better and louder?

■ is there any way at all to produce better video? again, would it help if the initial rendering was at (say) 720x405, 640x360 or even lower res, rather than 1920x1080?

 

these are just experimental ramblings...pretty sure this is going to be a case of Atari is already doing more than it should be able to?

 

would also be interest to hear what res @ndary video on page 7 was rendered at initially (ie prior to virtualdub being involved)

I can't offer any specific help.  I use linux utilities to do conversions.  

 

For video I'm at the mercy of mencoder's default settings and whatever video format converters are used.  I don't think the original resolution matters since the video gets scaled to 160x192 frames for the Atari.

 

For sound I generally tried to boost gain as far as I could before the utility sox created clipping, which it was kind enough to say when it was.  For non music I might accept a small amount of clipping just to keep the average audio level higher.   I didn't like that I have to crank the speakers' volume near max to get decent levels.  I inevitably forget that I have the volume cranked up only to be rudely reminded at my next Atari session.

 

For videos with dynamic audio I turned to audacity to edit the intermediate WAV file.  I can't recall which parameter(s) I used though to compress the sound a little (or a lot).

 

I lost interest in video conversions when I learned the movies don't display full color on either of the 2 old CRT TVs I use.  The only time I can see full color is when I use a scan converter and a VGA monitor.  My last VGA CRT is on  its way out.  The one VGA LCD is quite poor in comparison.  I have only NTSC gear.

 

For anyone who has an NTSC display/system that can show all 128 colors of these movie I recommend boosting saturation by 200% during the transcoding process.  

 

Such adjustment is not needed for conversions intended for PAL displays.

 

-SteveS

Edited August 18, 2019 by a8isa1

[+CharlieChaplin]

On 5/30/2013 at 7:36 AM, phaeron said:

This method of video display would be amenable to other kinds of displays besides mixed modes 9 + 11, but I haven't found another method that produces nearly as good of a picture. I tried experiments with 2-bit and 1-bit luma to see if the higher resolution would make up for the loss in color depth, but even with aggressive dithering the display looked much worse than having 4-bit hue and 4-bit luma.

 

Hello @phaeron,

 

thinking of silent movies and your fantastic A8 movieplayer...

 

a) would it be possible to play them back in greyscales only (e.g. Gr. 9 + Gr. 9 interleaved or HIP format) ? There is already a converted Laurel & Hardy video available and using 256 colour mode for a b+w movie is a little overkill (maybe we can therefore get higher resolution, e.g. with HIP format or similar).  Guess we would need another movieplayer then...

 

b) silent movies are quite old and normally do not run with 50 fps or 60 fps (e.g. with 16, 18 or 20 fps) - if one converts them for the A8 would it still be necessary to use 50fps (PAL) or 60 fps (NTSC) or would half the framerate (25/30 fps) be enough (assuming there were another movieplayer that works with 25/30 fps then) ?!?

 

c) silent movies do not have audio, so would it be possible to convert + playback the movies without audio on the A8 ? (Maybe we would also need another movieplayer then.)

I found a webpage of the university of Toronto with lots of self-created movies in the late 90s and early 2000s that do not have sound and it would be nice to still be able to convert them without having to add audio manually...    Wooden Monkey

(And many of these movies use just 30fps.)

 

I would love to convert an excerpt from a silent movie (feat. e.g. Charlie Chaplin, Buster Keaton, Harold Lloyd, Laurel & Hardy) or one of the Wooden Monkey movies to the A8...  

 

-----

 

In the meanwhile here are some PAL format movies for use with SIDE/SIDE-2, UNO-cart or AVG-cart that have been converted by B.Herale for me (afaik, the only movies from the wooden monkey page WITH audio).

 

WM_Movies1.zip

[ivop]

17 minutes ago, CharlieChaplin said:

b) silent movies are quite old and normally do not run with 50 fps or 60 fps (e.g. with 16, 18 or 20 fps)

True, but I think you would be hard pressed to find the original 16/18/20 fps sources online. Most of them are already "converted" to higher frame rates and are really badly done, people walking way too fast, et cetera. Remember them seeing on the telly back in the eighties. But I believe there are some better remastered versions now on DVD or BluRay. And even on YouTube.

 

Hard boiled ??? & ? ? ?  HMMMM

 

[phaeron]

Gr.9 only or 9/10 is possible, just requires changes to the PRIOR register values. The lower four bits are off-limits as they are also written to CHACTL and VSCROL, but the upper four bits only have meaning for PRIOR. I've considered using the fifth player to brighten up the chroma lines to reduce the hue shifts on NTSC but couldn't figure out how to do it without needing to drop back to narrow width.

 

The video must be full 50/60Hz. The player doesn't have a framebuffer, it streams the data directly from the IDE device to ANTIC without going through RAM. With audio, it isn't possible to repeat frames because the audio is interleaved with the video -- during the active region the sound samples are also read directly from the IDE port.

 

Without audio, it would be possible to repeat frames. It's still not ideal to do so, because the converter uses temporal dithering to improve the color precision. This only works at full frame rate and becomes too noticeable at lower frame rates.

 

[Poison]

I would like to ask why in every video i created is on begining terrible BEEP sound. But in audio file is silent at the begining of video. It was on SIDE II player and on AVG player too. Am I doing something wrong ? Or is it feature of player.

 

Anyway it is absolute EPIC !

 

thank you for help

[Peri Noid]

Is there a tool chain or description, how the films can be recoded using Linux tools only? FFMPEG, Mplayer/encoder or whatever else? I'm just finishing my first recoding done with Windows tools but I have to run them in VBox - I'd prefer to use it natively. Thanks!

[a8isa1]

11 hours ago, Peri Noid said:

Is there a tool chain or description, how the films can be recoded using Linux tools only? FFMPEG, Mplayer/encoder or whatever else? I'm just finishing my first recoding done with Windows tools but I have to run them in VBox - I'd prefer to use it natively. Thanks!

I don't have an automated shell script but I use the following as my guideline:

#NTSC video

mencoder -nosound -of rawvideo -ovc raw -vf hue=0:1.5,scale=157:190,expand=160:192,format=yv12,harddup,swapuv -sws 6 -ofps 59.9227 'robot_spy.mp4' -o robot_spy.raw

ffmpeg -i robot_spy.mp4 robot_spy.wav

sox robot_spy.wav -C 0.5 -c 1 -b 8 -r 15700 robot_spy.u8 gain -l 10

./encvideo60n <robot_spy.raw robot_spy.mov

./encaudio60n <robot_spy.u8 robot_spy.aud

./mux60n robot_spy.mov robot_spy.aud robot_spy.dd

 

# PAL video

#mencoder -nosound -of rawvideo -ovc raw -vf hue=0:1,scale=169:192,expand=160:192,format=yv12,harddup,swapuv -sws 6 -ofps 49.86 robot_spy.mp4 -o robot_spy-pal.raw

#### ffmpeg -i robot_spy.mp4 robot_spy.wav ; not need if NTSC video is already processed

#sox robot_spy.wav -C 0.5 -c 1 -b 8 -r 15558 robot_spy-pal.u8 gain -l 10

#./encvideo50n <robot_spy-pal.raw robot_spy-pal.mov

#./encaudio60n <robot_spy-pal.u8 robot_spy-pal.aud

#./mux50n robot_spy-pal.mov robot_spy-pal.aud robot_spy-pal.dd

 

I only have NTSC hardware so with the exception of doing some conversions for ndary years ago I only work with the NTSC standard.

 

Unfortunately I can't recall how I arrived at some of the parameters that I used in my script.  I'm terrible about taking notes. Below is what I do remember.

 

frame rates (-ofps switch of mencoder) and sample rates (-r switch of sox) are required by Phaeron's movie format based on video standard.

 

Mencoder switches

-vf hue=0:1.5 increases NTSC saturation.  IIRC every 2nd scanline hasn't color information so this is how I correct the washed out appearence.  TVs might not display all 128 colors.  My scan converter does.

 

-sws 6  (sofware scaler type).  6 represents luma bicubic / chroma bilinear.   Subjective but I thought this looked the best.  Feel free to experiment.

 

swapuv.  Phaeron found this for me.  U and V planes were generated in the wrong order by default.

 

-vf scale=157:190,expand=160,192:  This is a workaround.  160x192 is the rendered canvas size however when I scaled directly to this size I get a terrible high pitched whistle throughout the audio. Using these parameters leaves a blank line at both the top and the bottom.  Somehow this eliminates the whistle.  I don't understand why.  The 157x190 size makes Atari reproduce an original 4:3 aspect (on the 160x192 canvas).

 

I usually run the script, editing the file name the go back and play with the gain (-l switch of sox).    Sox will report how much clipping you get.  Depending the the video you might leave a some clipping and keep a higher average volume.  Clipping is more objectionable if it happens during music.   If the audio is too dynamic I usually use Audacity to either do compression or adjust amplitude.    There's a way to test the amount of clipping in Audacity but I've forgotten.  I can't remember if generate the .U8 file straight from Audacity or not.

 

Sorry I wish I remembered more of the process.

 

-SteveS

 

[EDIT]

p.s.  I found the following notes to myself 

 

NTSC > 160/190 = 0.8421 ; 190 instead of 192 because of "green line"  <-- This is the real reason why I changed the scaled size.  Did I accidently solve the *whistle* problem?  I don't remember.

PAL  > 160/240 = 0.6667

above ratio * video aspect ratio will yield the horizontal number for mencoder "-vf scale" (e.g "158:190") filter

 

use

mplayer -vo null -ao null -identify -frames 0 foo.avi

to obtain video's width and height
calculate original aspect ratio from those

 

when writing the CF card use dd's conv=direct,notrunc parameters

Edited August 7, 2020 by a8isa1

[SenorRossie]

I wrote a small shellscript to automate encoding videos. It has been published on github https://github.com/senorrossie/sh-avf-encode it was mentioned in this thread earlier, IIRC