emkay Posted July 5, 2012 Share Posted July 5, 2012 I would expect such a tool to run a lot lot faster. Or did you use a slow compiler / interpreter programming language? Compared to the first version, it is a lot quicker now. The thing is, instead of doing 10M evaluations, we now expect to do 100M evaluations! If it was speeded up more, we'd expect to do 200M evaluations! and so on....! As my latest pictures show, the tool works. But with every evaluation, it puts a slight error in. If you can put the import picture to the max details and colour limit , the importer can do a "perfect" fit in less than 20M evaluations. But an Atari native 4 colour image can put the converter into an unsolvable run. There will never be a perfect solution for all pictures thrown at it. The "perfect" solution is given by the restrictions of the A8. The limit cannot be exceeded. So there is a clear goal. Quote Link to comment Share on other sites More sharing options...
ilmenit Posted July 5, 2012 Author Share Posted July 5, 2012 (edited) But an Atari native 4 colour image can put the converter into an unsolvable run. That's a lie again... http://www.atariage....75#entry2549680 The "perfect" solution is given by the restrictions of the A8. The limit cannot be exceeded. So there is a clear goal. Yes master, we know that the perfect conversion can be easily solved with a lot of "just" and "simply". btw, could you finally answer this question and present the input and output picture, so anyone can do it? "You have 3 steps there - low colour (3:58), medium colour (4:46), full colour (6:35). What kind of colour reductions are applied there? Which tool/option/algorithm did you use for that?" Edited July 5, 2012 by ilmenit Quote Link to comment Share on other sites More sharing options...
emkay Posted July 5, 2012 Share Posted July 5, 2012 Yes master, we know that the perfect conversion can be easily solved with a lot of "just" and "simply". Cool.Now if you know that, how about removing the bugs and adding the automatic colour increasing? Quote Link to comment Share on other sites More sharing options...
ilmenit Posted July 5, 2012 Author Share Posted July 5, 2012 Yeah, sure... While I really want to make the converter better I won't waste my time on things that you can't prove to work. Quote Link to comment Share on other sites More sharing options...
emkay Posted July 5, 2012 Share Posted July 5, 2012 (edited) But an Atari native 4 colour image can put the converter into an unsolvable run. That's a lie again... Blahblahblah.... Don't call it a lie... If the picture is a 4 colour image with a slight different colour set, the converter doesn't get that and misses the automatic correction. This also belongs to parts of real colour pictures. The A8 has this 128 colours with fixed 8 brightnesses. There is nothing more to calculate to. Every slight variation of greens red and browns were to ignore , also slight brightness differences.... they simply do not exist in the range of the available 128 colours. Edited July 5, 2012 by emkay Quote Link to comment Share on other sites More sharing options...
emkay Posted July 5, 2012 Share Posted July 5, 2012 (edited) Yeah, sure... While I really want to make the converter better I won't waste my time on things that you can't prove to work. Hm.... You really seem not to understand, what I'm writing about. How about asking some runes for wisdom? As long as the colour handling is that wishy-washy , things won't get really faster. 1200 Million evaluations , and the lips don't get calculated right. from this: output.xex And, please don't talk again of some local optimum nonsense, because it isn't. Probably the colour gets correct calculated, but it doesn't get set there. Reminder: http://www.atariage.com/forums/topic/156160-quantizator/page__st__625#entry2551204 Edited July 5, 2012 by emkay Quote Link to comment Share on other sites More sharing options...
analmux Posted July 5, 2012 Share Posted July 5, 2012 Emkay, when will YOU write your gfx converter? Maybe you said several times that you won't write any atari programmes anymore, but a PC-tool would still be an option. 2 Quote Link to comment Share on other sites More sharing options...
ilmenit Posted July 5, 2012 Author Share Posted July 5, 2012 (edited) Hm.... You really seem not to understand, what I'm writing about. How about asking some runes for wisdom? I understand that you are a liar and ignorant, who prefer to laugh at others and troll instead of helping. If you are not a liar then why it is so problematic for you to show everyone how simple it is to get from this picture: to the one you have shown on the movie at 3:56? http://www.youtube.com/watch?v=s1gLy3ZBkoc&t=3m56s If you won't do that I will continue to say that you are a liar, who lies how shitty the tool is when everything is SO simple. 1200 Million evaluations , and the lips don't get calculated right. from this: Here is a line (172) from the output.png-dst.png where the lips are: It has 10 colours. If you are so smart then show us the kernel program for just this single line that covers the colors perfectly (all the 10 colours). It does not even have to fit into colour changes and sprite repositioning needed for the previous and next lines. And, please don't talk again of some local optimum nonsense, because it isn't. You don't understand what the hill climbing is and you don't understand what the local optimimum is. Go and learn. The others gave you links for reading. Probably the colour gets correct calculated, but it doesn't get set there. Calculating a color is trivial. To set it in the right place without screwing the rest of the picture is hard. Edited July 5, 2012 by ilmenit 2 Quote Link to comment Share on other sites More sharing options...
Heaven/TQA Posted July 5, 2012 Share Posted July 5, 2012 Emkay, when will YOU write your gfx converter? Maybe you said several times that you won't write any atari programmes anymore, but a PC-tool would still be an option. Because he is no programmer that's why he will not touch any code.... doesn't matter which plattform. but guys, as I mentioned earlier, try to get on a level to help each other. So emkay show how you convert stuff step by step and the guys will have a look at the converter if there is a unknown bug or if some of the already implemented algorithms can be approved. just my 2 cents. 1 Quote Link to comment Share on other sites More sharing options...
emkay Posted July 5, 2012 Share Posted July 5, 2012 Emkay, when will YOU write your gfx converter? Maybe you said several times that you won't write any atari programmes anymore, but a PC-tool would still be an option. Because he is no programmer that's why he will not touch any code.... doesn't matter which plattform. but guys, as I mentioned earlier, try to get on a level to help each other. So emkay show how you convert stuff step by step and the guys will have a look at the converter if there is a unknown bug or if some of the already implemented algorithms can be approved. just my 2 cents. Not sure, where the problem is, to use any tool.... Gimp , or else.... for converting the picture into indexed colours. On the other hand, this "short" Video needed 4h for being uploaded. Creating an image with reduced colour , anyhow, is in the creator's eye and adjustment. The video explains enough.... Quote Link to comment Share on other sites More sharing options...
ilmenit Posted July 5, 2012 Author Share Posted July 5, 2012 (edited) Creating an image with reduced colour , anyhow, is in the creator's eye and adjustment. This is the point. For a long time you propose 2 things to be done automatically by the tool: 1. To reduce colours of the picture to keep details where needed 2. To iteratively increase number of colours to put more details into the process I want you to understand that even the first operation can't be done as you wish. It depends totally on a human eye to decide where details of the picture should be. This is what you did with the picture on the movie - you left colours on the ladies, because you want to remove banding on that area of the picture. Any existing colour reduction algorithm would prefer the sky and the sea, and then iterative colour increase only make things worse. 4 colours 8 colours Instead of calling the tool half-baked and laughing at people maybe you should write some tutorial how do you put more details in the picture? That would bring something good to the others. Edited July 5, 2012 by ilmenit 4 Quote Link to comment Share on other sites More sharing options...
potatohead Posted July 5, 2012 Share Posted July 5, 2012 I'm not sure I would take that kind of shit on one of my projects. ilmenit, you have the paitence of Job. Emkay, why not simply show the steps? [start picture] Emkay does A, B, C [next picture] Emkay does A, B, C [final picture] It's not hard. Personally, I think he just colored them to communicate desired result, and "it's possible on A8, so make it happen" more than any real process. 3 Quote Link to comment Share on other sites More sharing options...
emkay Posted July 5, 2012 Share Posted July 5, 2012 (edited) I'm not sure I would take that kind of shit on one of my projects. Creating an image with reduced colour , anyhow, is in the creator's eye and adjustment. This is the point. For a long time you propose 2 things to be done automatically by the tool: 1. To reduce colours of the picture to keep details where needed 2. To iteratively increase number of colours to put more details into the process I want you to understand that even the first operation can't be done as you wish. It depends totally on a human eye to decide where details of the picture should be. This is what you did with the picture on the movie - you left colours on the ladies, because you want to remove banding on that area of the picture. Any existing colour reduction algorithm would prefer the sky and the sea, and then iterative colour increase only make things worse. 4 colours 8 colours Instead of calling the tool half-baked and laughing at people maybe you should write some tutorial how do you put more details in the picture? That would bring something good to the others. Could you please stop to talk to each other with your totally wrong thoughts? I know, people put their own thoughts into other's ... so I'm still not angry from your weird attacks..... Quick import to the A8 "fitting format" , simply dropping the "original" through your tool, then using the dst picture, and adjust the resolution... no additional colour calculation. merswin.bmp Simple setting in Gimp (converting tool to create indexed palettes ... alike which) to 8 colour indexed palette... You may use Floyd-Steinberg calculation for that. merswin8.bmp Edited July 5, 2012 by emkay Quote Link to comment Share on other sites More sharing options...
emkay Posted July 5, 2012 Share Posted July 5, 2012 (edited) What really would help, was to find the causes for the two special "Errors". That colour reduce is only a way to handle the slowing down issues... Error 1 shows coloured pixels where they weren't needed. They weren't in the source, and they weren't in the dst ... Error 2 shows a "band" in the center of a coloured region... which is very common in the pictures. That band isn't there , because the A8 cannot handle that "place"... it's there due to some calculation issue. Edit: at 225 mil, the "error 2" has gone. Edited July 5, 2012 by emkay Quote Link to comment Share on other sites More sharing options...
José Pereira Posted July 5, 2012 Share Posted July 5, 2012 (edited) Can I interrupt you and ask for a favour? I re-painted using PAINT the Cybernoid2 Loading screen to other colours that would fit better on a Cybernoid2 A8 version and just because you seems best and better than me, can you try this picture and put here the results? Here we have the '.png' using P.C. PAINT colours but when I tried to Load it into G2F using Laoo Pallete I get this (of course there's colour clash in a direct ColourMap VBXE version, but I think it would look better with Rasta's Converter and on a stock XL/XE Machine) and are, more or less, the colours I am expecting to get on A8: A help would be very much appreciated. Thanks. Edited July 5, 2012 by José Pereira Quote Link to comment Share on other sites More sharing options...
ilmenit Posted July 5, 2012 Author Share Posted July 5, 2012 What really would help, was to find the causes for the two special "Errors". I see my efforts to explain were pointless... Emkay goes to the ignored users list. 1 Quote Link to comment Share on other sites More sharing options...
emkay Posted July 5, 2012 Share Posted July 5, 2012 What really would help, was to find the causes for the two special "Errors". I see my efforts to explain were pointless... Emkay goes to the ignored users list. Cool. So that button do you find? Wow. But the brain isn't big enough to understand the simplest things to get the wanted result. Instead of stating, you don't really know what you do, you always put the agression toward me. Very nice. That darker pixels have - in no way- something to do with that basic hill climbing algorithm. There is a calculation bug happening, but emkay is the bad boy .... lol Quote Link to comment Share on other sites More sharing options...
emkay Posted July 5, 2012 Share Posted July 5, 2012 In real life, Clearasil may help.... The darker pixel around the brighter ones were artefacts from my "step by step" colour enhancing. Now they stay for 250 mil evaluations. But they were to remove without any calculaton of any "hill climbing". The error is in the bitmap and easy to compare, and the pixel simply could be removed. That "checking" is missing, and explains aswell, why the calculations need that long. The details get sharper when the random hit reaches a valueable pixel. Vice versa, false details force the whole routine to calculate wrong places. Knowing the exact calculation positions, speeds the whole thing up. But Sir Ilmenit doesn't get it. Quote Link to comment Share on other sites More sharing options...
emkay Posted July 5, 2012 Share Posted July 5, 2012 It's not hard. It's not hard.... It's useless, as it has pointed again. It costs time and gets no copy. Letting the generator run, the PC does in the background.... Quote Link to comment Share on other sites More sharing options...
+Stephen Posted July 5, 2012 Share Posted July 5, 2012 In real life, Clearasil may help.... The darker pixel around the brighter ones were artefacts from my "step by step" colour enhancing. Now they stay for 250 mil evaluations. But they were to remove without any calculaton of any "hill climbing". The error is in the bitmap and easy to compare, and the pixel simply could be removed. That "checking" is missing, and explains aswell, why the calculations need that long. The details get sharper when the random hit reaches a valueable pixel. Vice versa, false details force the whole routine to calculate wrong places. Knowing the exact calculation positions, speeds the whole thing up. But Sir Ilmenit doesn't get it. Is that dark line a playfield coloured pixel or is it possibly a quad-width player which may be needed elsewhere? Quote Link to comment Share on other sites More sharing options...
emkay Posted July 5, 2012 Share Posted July 5, 2012 Really, if he hadn't written the following, I even wouldn't have even posted the additional merswin pictures. Yeah, sure... While I really want to make the converter better I won't waste my time on things that you can't prove to work. If Ilmenit really likes to make the converter better, why does he stand on that time consuming routines? If the "RastaConverter" is just some playground for "learning about the hill climbing theory" and the A8 conversions is a side-product, he shouldn't name it "converter". Quote Link to comment Share on other sites More sharing options...
emkay Posted July 5, 2012 Share Posted July 5, 2012 Is that dark line a playfield coloured pixel or is it possibly a quad-width player which may be needed elsewhere? That doesn't matter. If it is a Quad width player, it could be immediately repositioned to where it would be needed. And that would happen, if that check wouldn't need more than the time 250M of evaluations. Meanwhile the pixel were removed... btw. Quote Link to comment Share on other sites More sharing options...
emkay Posted July 5, 2012 Share Posted July 5, 2012 Really. No on is expecting miracles. demo.xex But things really could get faster.... Quote Link to comment Share on other sites More sharing options...
Irgendwer Posted July 5, 2012 Share Posted July 5, 2012 That darker pixels have - in no way- something to do with that basic hill climbing algorithm. There is a calculation bug happening, but emkay is the bad boy .... lol http://en.wikipedia.org/wiki/Dunning%E2%80%93Kruger_effect 1 Quote Link to comment Share on other sites More sharing options...
Irgendwer Posted July 5, 2012 Share Posted July 5, 2012 That doesn't matter. If it is a Quad width player, it could be immediately repositioned to where it would be needed. And that would happen, if that check wouldn't need more than the time 250M of evaluations. http://en.wikipedia.org/wiki/Combinatorial_explosion 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.