Quantizator

[1NG]

is there a way to have a ASM-Source file? In theory the generator assembles via MADS anyway the .xex? Sorry for not having checked the output files manually.

The output is asssembler source code. The batch in the generator folder makes a xex of that.

Edited May 2, 2012 by 1NG

[Heaven/TQA]

and which file?

[andym00]

Just open the .rp file..

[1NG]

and which file?

 

BTW: rp = raster program

And for all people that don´t know computers very well (One argument by Apple is to not have to)

Easiest way to look into a file: Use F3 on Total Commander. If it contains LDA or STA it is the right one :-) sorry, could not resist.

[frogstar_robot]

 

Or you could try this build.

 

Any chance you can post the source for it? Ivop's Linux patch worked a treat for me and I'd like to try a native version of your new branch.

 

Thanks.

[Heaven/TQA]

and which file?

 

BTW: rp = raster program

And for all people that don´t know computers very well (One argument by Apple is to not have to)

Easiest way to look into a file: Use F3 on Total Commander. If it contains LDA or STA it is the right one :-) sorry, could not resist.

 

Don't worry. as Manager I am first asking the right persons before I act myself

[Marek Konopka]

Any chance you can post the source for it? Ivop's Linux patch worked a treat for me and I'd like to try a native version of your new branch.

 

I intend to modify the source further and improve the quality of the algorithm before I post the source. Need couple of days for that.

Edited May 2, 2012 by Marek Konopka

[Heaven/TQA]

i run some test pics on an quadcore with more than 3000k evaluations. and it is amazing what kind of results you got... let's see where this kind of tool will lead to.

[AtariNerd]

 

Tried playing around with the proportions and resampling the original Amiga image to get a little better detail (there was a great deal of empty screen between the characters, which was probably better from a dramatic staging effect, but a bit problematic when you're dealing with just 160 pixels...) , which was succesful, but still had a few areas where there was color clash, which the resampling doesn't help with much, because it can create pseudocolor effects. It wasn't any worse than the previous attemp, though and overall a better result, until you compare it to what the original revamped image looked like. :

 

 

 

Oh, well.

 

Will try re-running it through the latest patched version; looks like it might swap out a few colors or do a few color/value conversions and give a different balance to the image. This is such a fun toy/tool.

sh_amiga_mod.xex

Edited May 2, 2012 by AtariNerd

[1NG]

The base picture contain 51 atari colors. The achieved pic above is impressive for an automatically generated pic!

There is still some distortion though.

 

The theoretical best picture with atari colors looks like that:

 

For comparison:

a)And a flickering CIN-Picture of that:

the colors are a bit washed out. (CIN has only 4 shades and limited color intensity)

 

b)The best 4 color pic looks like that:

You see good colors but due to dithering a lot loss of sharpness.

 

 

Maybe it is impossible to get a better picture with a raster program, but I totally agree that it is a fun toy/tool

Edited May 2, 2012 by 1NG

[+MrFish]

The base picture contain 51 atari colors. The achieved pic is impressive for an automatically generated pic! There is still some distortion though.

 

The theoretical best picture with atari colors looks like that:

 

 

For comparison:

a)And a flickering CIN-Picture of that:

b)The best 4 color pic looks like that:

You see good colors but due to dithering a lot loss of sharpness.

 

 

Maybe it is impossible to get a better picture with a raster program, but I totally agree that it is a fun toy/tool

 

(I don´t know why the heck the 4c picture is smaller in here.)

 

That looks almost impossible. I'm impressed.

[AtariNerd]

Actually, that second image was the source converted to Atari colors and 160 pixels that the program generated, theoutput.png-src.png file, so that is the theoretical best image - I have lost my original image, so I just grabbed and used that one, instead. :/ In the executable, the displayed image has about 60 colors, if I remember correctly, so some color reduction may have helped..

Edited May 2, 2012 by AtariNerd

[emkay]

Maybe it is impossible to get a better picture with a raster program, but I totally agree that it is a fun toy/tool

 

It's possibly in the "upgrades" of the tools. In early G2F times I made an own picture (collage) it was actually the 1st conversion... Today it could look better ...

 

[+MrFish]

The base picture contain 51 atari colors. The achieved pic above is impressive for an automatically generated pic!

There is still some distortion though.

 

The theoretical best picture with atari colors looks like that:

 

For comparison:

a)And a flickering CIN-Picture of that:

the colors are a bit washed out. (CIN has only 4 shades and limited color intensity)

 

b)The best 4 color pic looks like that:

You see good colors but due to dithering a lot loss of sharpness.

 

 

Maybe it is impossible to get a better picture with a raster program, but I totally agree that it is a fun toy/tool

 

OK, I jumped the gun. I guess we're not there yet.

 

It's still an impressive tool. I think MK's idea of color reduction before processing seems like it would be helpful. I guess I need to try it out myself.

[ivop]

Color reduction is not always helpful as a preprocessing stage. Otherwise you could just reduce the amount of colors on each line to four (or six or eight or whatever) and be done with it. Sometimes several colors will map to the same Atari 8-bit color (which is why using different destination palette files results in different solutions). Increasing contrast and/or changing hues of certain colors can also be beneficial. If color and contrast stand out more, chances are they will map to different 8-bit colors and have a higher probability to be kept in the end (instead of being somewhat approximated by dithering with other colors).

[emkay]

 

It's still an impressive tool. I think MK's idea of color reduction before processing seems like it would be helpful. I guess I need to try it out myself.

 

Colour reduction is not the correct word. Colour-Room reduction fits better. All that helps to remove dither pixels, keeps resources free to set more image-details.

[emkay]

Have a look at this picture:

 

 

Without dither, many "typical" Atari colours get missed. In the sky for example.

 

 

 

To have more colours in the sky, dither has to be activated...

 

 

 

But this results in adding colour where no additional colour was needed. So the image looks weird and the imagedetails never get finalized.

[emkay]

So a "brightness dither only with the next available Atari colour" would help much there.

 

[emkay]

A possibility:

 

1st create a b/w image

 

 

 

and put a crosssampling with the "next possible" colour over that.

 

 

 

[+MrFish]

Color reduction is not always helpful as a preprocessing stage. Otherwise you could just reduce the amount of colors on each line to four (or six or eight or whatever) and be done with it.

 

I have no plans of going through any piece of artwork and reducing/counting the colors line by line. I think that would be an arduous task that defeats the purpose of the tool. What I mean by reducing the colors is to bring the total color count down with the image in my control and keep a general idea of how the colors are distributed horizontally. Whether this will work or not remains to be seen, as I'm just getting started with it.

 

Increasing contrast and/or changing hues of certain colors can also be beneficial. If color and contrast stand out more, chances are they will map to different 8-bit colors and have a higher probability to be kept in the end (instead of being somewhat approximated by dithering with other colors).

 

Using brightness and contrast adjustments are good methods, in fact using the contrast essentially amounts to a form of color reduction. I've been using this technique for a long time and it's exactly one of the methods that I plan on employing here.

[+MrFish]

It's still an impressive tool. I think MK's idea of color reduction before processing seems like it would be helpful. I guess I need to try it out myself.

 

Colour reduction is not the correct word. Colour-Room reduction fits better. All that helps to remove dither pixels, keeps resources free to set more image-details.

 

Maybe I misunderstood your initial post about it. I'll go back and read it after I've had some sleep and see if I understand what your saying here.

[+MrFish]

It's still an impressive tool. I think MK's idea of color reduction before processing seems like it would be helpful. I guess I need to try it out myself.

 

Colour reduction is not the correct word. Colour-Room reduction fits better. All that helps to remove dither pixels, keeps resources free to set more image-details.

 

Maybe I misunderstood your initial post about it. I'll go back and read it after I've had some sleep and see if I understand what your saying here.

 

OK, I went back to your original post(s) and also read through the thread a little more thoroughly -- I haven't been following the details, just checking the output that people are getting. I guess "colour-room" is a little more descriptive, but you're still reducing the amount colors, so I wouldn't say using the term "color reduction" is incorrect.

 

Anyway, I've played around with this enough to come up with a decent approach. I'll post some results once I have something worth posting...

[emkay]

I guess "colour-room" is a little more descriptive, but you're still reducing the amount colors, so I wouldn't say using the term "color reduction" is incorrect.

 

 

Right now, the tool is just a neutral converter. Using the "available" colours by now.

 

Have a look at post 192 , the last picture. The dithering produces a grey colour in the walls. Without dithering, the available colour is consistent and fits well. As you "know" that, you could tell the tool, don't change that colour in that range. So it leaves resources free to put more details on the Landing pad with the Ornithopter.

At the end there is no colour reduction..... in the original there hasn't been "grey" . A colour gain is possible also. Because missed colours of the original could be used better.

 

 

[emkay]

Another example:

 

 

You see on the "destination" that grey is mixed everywhere. The Bag is originally brown with some grey in there.

The "destination" view shows a huge mixing with grey.

And at the end, you have a grey bag with some browns.

Similary things happen with the sky and the grass and the stones....

 

It would help to build a "rule" into the converter, to use grey only, if there really is a grey in the original.

 

.... as the real colour room is much bigger than from the A8 palette, grey shifts the look already. Consistent colours would shift the look also, but in a nicer way, as it gets more colourful and uses less transitions.

Edited May 4, 2012 by emkay

[matosimi]

It would help to build a "rule" into the converter, to use grey only, if there really is a grey in the original.

 

Dude, destination image is approximated to atari palette file. There is algorithm used to find most similar color from original image in selected atari palette. Grey is just evaluated as best match. Tell me how would you decide not to use evaluated color but some other one which could be better for your eye? - it's not easy task to find proper solution nor code it.

 

I can imagine that after destination image is created, there could be 2nd conversion which would search for greys (reminded me X-files ) and replace them somehow with neighboring non-grey colors.

But this have to be also crosschecked with colours from source image to retain greys which have to be greys... -and this is exactly the problem. Because you have to state deviation which would tell algorithm how far from exact grey (RGB = XYZ, where X=Y=Z) it can go and still be evaluated as grey.

Hope u can understand what I mean...

 

So, instead of this masturbation, I recommend you to experimentally add more saturation to your input picture and see if you can get result u like .