emkay Posted June 3, 2012 Share Posted June 3, 2012 I'm pretty sure, the images could be converted in some seconds. The converter does steps that complex things more, and to handle them using more time. You know this picture Using standard import, you get either this: or this: In both cases, the converter adds colours, that don't get handled with the importing filter... The picture never gets really finished. The way to the perfect solution has to be changed a bit .... Quote Link to comment Share on other sites More sharing options...
emkay Posted June 3, 2012 Share Posted June 3, 2012 You could do a conversion, using a real iteration of the image, cross checking the available colours, plus resolution, plus possible changes per line were reached. Quote Link to comment Share on other sites More sharing options...
emkay Posted June 3, 2012 Share Posted June 3, 2012 As you get a 1st indexed picture with a defined palette, you could do a 1st adjustment of the palette, on the direct used colours... Which gives a result like this in less than a minute: 1 Quote Link to comment Share on other sites More sharing options...
emkay Posted June 3, 2012 Share Posted June 3, 2012 The next step explains itself. Having this picture, and the original picture, you could compare the ranges of the picture, whether a colour takes more to disturb the picture showing a line, or a pixel. And to compare, whether colours exceed the limit of the possible changes, or changes were still available for adding some details there... If the image is "disturbed" like with the occuring lines, the colours should be exchanged and reduce by one. And another colour gets used there. A single pixel with a wrong colour isn't as much disturbing than a huge scanline "error". Quote Link to comment Share on other sites More sharing options...
Rybags Posted June 3, 2012 Share Posted June 3, 2012 That picture would probably go better shifted a bit to the right - allowing more cycles for changes. Quote Link to comment Share on other sites More sharing options...
emkay Posted June 3, 2012 Share Posted June 3, 2012 Or to put it in short words: The picture can get prepared until it directly fits into the code that has to be used to show it on the A8. Quote Link to comment Share on other sites More sharing options...
emkay Posted June 3, 2012 Share Posted June 3, 2012 That picture would probably go better shifted a bit to the right - allowing more cycles for changes. You cannot move all images to the right... robbi26m.xex 26M evals. 1 Quote Link to comment Share on other sites More sharing options...
emkay Posted June 3, 2012 Share Posted June 3, 2012 Really. As the converter is doing that much, it really could do the rest aswell Preprocessing of the available colours, helps to remove the "lines" and unnecesary dither pixels.... DonBlech.xex 1 Quote Link to comment Share on other sites More sharing options...
emkay Posted June 3, 2012 Share Posted June 3, 2012 (edited) ... and you don't need to wait for 100 millions of evaluations... , which costs the most time. Edited June 3, 2012 by emkay Quote Link to comment Share on other sites More sharing options...
snicklin Posted June 3, 2012 Share Posted June 3, 2012 Why is it calculated to 320 pixels width? Why not 160? Is there some added accuracy? Quote Link to comment Share on other sites More sharing options...
emkay Posted June 3, 2012 Share Posted June 3, 2012 Why is it calculated to 320 pixels width? Why not 160? Is there some added accuracy? If it's at 320 pixel, the A8 could not show it. The images were set to 320 width, just for the correct aspect ratio. Quote Link to comment Share on other sites More sharing options...
snicklin Posted June 3, 2012 Share Posted June 3, 2012 Why is it calculated to 320 pixels width? Why not 160? Is there some added accuracy? If it's at 320 pixel, the A8 could not show it. The images were set to 320 width, just for the correct aspect ratio. Maybe I'm being daft, I still don't get it.... Yes, 320 pixels is too many pixels in width. So why not calculate it to 160 width x whatever is set for the height? The thing is, if you calculate it to 320 pixels wide, aren't you doing twice as much work as you need? Now if this breaks the aspect ratio, so be it, because as soon as you use the generator, it will break the aspect ratio anyway in converting it to 160 x H pixels. Sorry if I am missing the point here.... Quote Link to comment Share on other sites More sharing options...
w1k Posted June 3, 2012 Share Posted June 3, 2012 how i can make picture without border? /noborder - no works Quote Link to comment Share on other sites More sharing options...
Xuel Posted June 3, 2012 Share Posted June 3, 2012 how i can make picture without border? /noborder - no works You could change this line in no_name.asq: mva #$ff sizem to this: mva #$00 sizem Quote Link to comment Share on other sites More sharing options...
w1k Posted June 3, 2012 Share Posted June 3, 2012 i dont see any "mva #$ff sizem" code in that file :/ Quote Link to comment Share on other sites More sharing options...
+CharlieChaplin Posted June 3, 2012 Share Posted June 3, 2012 Simple, how does a 320 x 200 picture look on PC or other computers ?!? ----------------- Picture here ----------------- and how does a 160x200 picture look on the PC or other computers ?!? --------- Picture here Picture here --------- On the A8 however, no matter if you use 320x200, 160x200 or 80x200 a picture will always look like this: ---------------- Picture here ---------------- Thats why I always use 320x200 (or 320x192) pixels reolution when converting a picture from PC or other computers to the A8. -Andreas Koch. Quote Link to comment Share on other sites More sharing options...
Xuel Posted June 3, 2012 Share Posted June 3, 2012 i dont see any "mva #$ff sizem" code in that file :/ Can you try using Beta4? Quote Link to comment Share on other sites More sharing options...
w1k Posted June 3, 2012 Share Posted June 3, 2012 (edited) deleted Edited June 3, 2012 by w1k Quote Link to comment Share on other sites More sharing options...
snicklin Posted June 3, 2012 Share Posted June 3, 2012 ... Thats why I always use 320x200 (or 320x192) pixels reolution when converting a picture from PC or other computers to the A8. -Andreas Koch. OK, I see your point there when talking about input files. However, using a height of 192 as an example, why don't we just take 320x192 as a PC picture as the input, then convert it to 160x192 for the perfect destination picture and then calculate towards the 160x192 picture using a 160x192 picture? i.e. Input=320x192 Intermediate (output pic)=160x192 Perfect Atari Pic=160x192 Then the generator works off the Intermediate pic. Currently the generator converts from 320x192 to 160x192. Why do it afterwards? Quote Link to comment Share on other sites More sharing options...
emkay Posted June 3, 2012 Share Posted June 3, 2012 ... Thats why I always use 320x200 (or 320x192) pixels reolution when converting a picture from PC or other computers to the A8. -Andreas Koch. OK, I see your point there when talking about input files. However, using a height of 192 as an example, why don't we just take 320x192 as a PC picture as the input, then convert it to 160x192 for the perfect destination picture and then calculate towards the 160x192 picture using a 160x192 picture? i.e. Input=320x192 Intermediate (output pic)=160x192 Perfect Atari Pic=160x192 Then the generator works off the Intermediate pic. Currently the generator converts from 320x192 to 160x192. Why do it afterwards? It's only for the correct aspect ratio.... just a pixel re-use. Quote Link to comment Share on other sites More sharing options...
emkay Posted June 3, 2012 Share Posted June 3, 2012 Seems I found an error... This should be: This "is" : error.xex Quote Link to comment Share on other sites More sharing options...
emkay Posted June 3, 2012 Share Posted June 3, 2012 Possibly it happened with changing the revision. As a pink colour was interfering the picture, I removed it from the palette, and did a restart... some days back. Hm... Quote Link to comment Share on other sites More sharing options...
ilmenit Posted June 4, 2012 Author Share Posted June 4, 2012 (edited) I'm pretty sure, the images could be converted in some seconds. The converter does steps that complex things more, and to handle them using more time. [...] In both cases, the converter adds colours, that don't get handled with the importing filter... The picture never gets really finished. The way to the perfect solution has to be changed a bit .... Sorry emkay, but I do not understand. Could you describe (in points) what exactly do you recommend or what you did with those pictures, please? The default /init is 'random', which gives the best results for complex pictures. With random initialization the optimization algorithm usually covers the pictures with sprites best. For simple pictures you may try /init=smart or /init=less or /init=empty @snicklin - RastaConverter converts every input picture to width=160 pixels (/filter parameter) Edited June 4, 2012 by ilmenit Quote Link to comment Share on other sites More sharing options...
ilmenit Posted June 4, 2012 Author Share Posted June 4, 2012 I'll give dither_val a try! I really like the results of ciede but, man, it is slow. With only 200000 evaluations it may be worth to try /init=smart. With /distance=ciede the slowness at the beginning is caused by default /init=random. Quote Link to comment Share on other sites More sharing options...
emkay Posted June 4, 2012 Share Posted June 4, 2012 I'm pretty sure, the images could be converted in some seconds. The converter does steps that complex things more, and to handle them using more time. [...] In both cases, the converter adds colours, that don't get handled with the importing filter... The picture never gets really finished. The way to the perfect solution has to be changed a bit .... Sorry emkay, but I do not understand. Could you describe (in points) what exactly do you recommend or what you did with those pictures, please? Actually, I have no idea how to explain it else. It's already a step by step explanation. A short list: Put the picture through the converter, to have the 160x240 pixel. Check the available colours per line. Check the available Atari colours. Drop the colours down to a countable use, for the available colour changes and usable colours.... per line Put the "original" picture again through the converter with the aimed "fitting" colours and use dithering for that. Then no dither again! As you have an 160x240 pixel picture with a limited count of colours, now an extra filter checks the used colours in the picture and exchanges them with the next possible Atari colour (from the Laoo palette for example)... Then run it through the converter. If all calculations were done before, the import to "Atari-Code" is rather fast. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.