Jump to content
IGNORED

Images generated by RastaConverter


Philsan

Recommended Posts

Another recently finished conversion and one of many long runners I'm just wrapping up to post. :) 

Thought I'd try and duplicate this stylised comic book style source image as accurately as possible colour wise. (I am often flexible with some of my conversion's colours which can often intentionally differ from the source if the output image looks good. (My Samaurai Tattoo conversions being good example of that).

 

Few line errors down sides of face and in the background. Rightside of face shading could have done with more time. Really pleased with the green text definition and the overall high colour count.

 

Nebula (of Guardians of the Galaxy universe)

 

3.6 bill evaluation

94 unique colours

Filename: Beeblebrox_Nebularcomicsm1_13.xex

(Altirra image with PAL high artifacting enabled)

image.thumb.png.c9b8bc502ca37494e3809aa1a33775d0.png

 

Beeblebrox_Nebularcomicsm1_13.xex

 

Source file was reduced to 128 colours  RGB lines and saturation, contract and gamma tweaked/image resized to 320*240 in Irfanview, then processed in RC at 320*240 after a few more RC settings were tweaked

Settings were: Filter = bicubic; pal = altirra; color dis = yuv; initial state = smart; no of solutions = 20000; Threads = 9; preprocess = yuv; brightness = 0; contrast = 39; gamma = 0.8; dither = none; dither strength  = n/a; randomness = n/a; mask value 15
 

Here is the conversion in progress - mask applied to quite a bit of the image

(- colour match pretty close to the source):

image.thumb.png.e3ba6c6086017f431f1f5991cc581fd4.png

 

..this is what it looks like on my B&O MX4000 CRT running off my 800 via S-video,

(again as per previous post colours differ slightly in the image running on this CRT

off the stock 800, (ie some pinks look purple, purples look blue, etc)):

image.thumb.png.7efe90f66f6bdc7ad59dfe299a2d60a9.png

 

...and here on my 10" CRT (800XL UAV rD) where colour match closer to the Altirra screengrab

(ignore light reflection left of screen and overexposed shot btw):

image.thumb.png.7a08efd58d0d708034bc8c7b76f8b2f6.png

 

Edited by Beeblebrox
  • Like 1
Link to comment
Share on other sites

On 1/19/2022 at 4:58 PM, Beeblebrox said:

I decided I wanted to try an convert a series of lighthouse images. This was one of the more successful ones, a moody dawn - (I forgot to note down the location - somewhere in the States I think).

 

Lighthouse Lights

 

401 mill evaluations

69 unique colours

Filename: Beeblebrox_Lighthouselightssm1_04.xex

(Altirra image with PAL high artifacting enabled)

image.thumb.png.76bd7b1af52985f4b50ddb7f5eede761.png

Beeblebrox_Lighthouselightssm1_04.xex 22.11 kB · 5 downloads

 

 

Dusk, actually, on the east shore of Lake Michigan in Grand Haven, ten miles from where I grew up. Thanks for posting!

Edited by ClausB
  • Like 2
  • Thanks 1
Link to comment
Share on other sites

Two more long runners finished.:D Geralt and Ciri artwork from the Witcher 3 game processed as two separate images. 

Some line errors and banding (in particular in the Geralt image - annoying brown banding). Masks applied to character's profiles and not backgrounds. (Not too concerned about the line errors in the background - background mainly there to give context to the eerie green glowon character's skin tones:thumbsup:).

 

In both cases: 

Source file was reduced to 128 colours  RGB lines and saturation, contract and gamma tweaked/image resized to 320*240 in Irfanview, then processed in RC at 320*240 after a few more RC settings were tweaked

Settings used: Filter = bilinear; pal = altirra; color dis = yuv; initial state = smart; no of solutions = 20000; Threads = 9; preprocess = yuv; brightness = 3; contrast = 28; gamma = 1.1; dither = floyd; dither strength  = 1.0; Mask value = 20

 

*****

 

Ciri image (my fav of the two :grin:):

 

3.7 bill evaluations

58 unique colours

Filename: Beeblebrox_Cirisidesm1_08_fixed.xex

(Altirra image with PAL high artifacting enabled)

image.thumb.png.f78afe71bb4a34cb81589c4333c5b51a.png

Beeblebrox_Cirisidesm1_08_fixed.xex

 

*****

 

Geralt image:

 

3 bill evaluations

60 unique colours

Filename: Beeblebrox_Witcherprofilesm1_12.xex

(Altirra image with PAL high artifacting enabled)

image.thumb.png.af1ef6152d5a40254828598d07dc2878.png

Beeblebrox_Witcherprofilesm1_12.xex

 

Here is the original Google image from which I split up and cropped the source images:

image.thumb.png.7ce124f82703f24f22b103db7812ca46.png

 

 

Ciri on my B&O MX4000 CRT (800 S-video) - really good match for colours IMHO:

image.thumb.png.442118bbab57a1095bbeb3e084a36abd.png

 

and Geralt on the same:

image.thumb.png.e72f3e058988142e3ec5b700d5889ab4.png

 

 

 

 

 

 

Edited by Beeblebrox
  • Like 4
Link to comment
Share on other sites

Another lighthouse, ( @a8isa1  this one's for u :grin:).

 

Sylvan Lake lighthouse, Sylvan Lake, Alberta in Canada. 

 

Quite a moody pic, (post or pre- storm - who knows?:P)

Converted fairly well. Annoying banding in the upper clouds I just couldn't shift but it kinda blends in enough not to be too obvious.

 

I used Sheddy's great PM line artifact removal tool to successfully get rid of a major white line right in the middle of the image so all good.:D

 

1.7 Bill evaluations

74 unique colours

Filename: Beeblebtox_Sylvanlighthousesm1_09_fixed.xex

(Altirra image with PAL high artifacting enabled)

image.thumb.png.c404b204133fdd4cf20e5c81a451f2fe.png

Beeblebrox_Sylvanlighthousesm1_09_fixed.xex

 

Source file was reduced to 128 colours  RGB lines and saturation, contract and gamma tweaked/image resized to 320*240 in Irfanview, then processed in RC at 320*240 after a few more RC settings were tweaked

Settings used: Filter = bilinear; pal = altirra; color dis = yuv; initial state = smart; no of solutions = 20000; Threads = 9; preprocess = yuv; brightness = 2; contrast = 39; gamma = 1; dither = none; dither strength  = n/a; randomness = n/a; mask value 20

 

on my B&O MX4000 CRT (800 S-video):

image.thumb.png.0a39102666f1f5898e40497b4de5c860.png

 

 

and on my 10" CRT (800Xl UAVrD, S-video):

image.thumb.png.8e7f099d49f90508bb72a836810556f3.png

Edited by Beeblebrox
  • Like 2
Link to comment
Share on other sites

@a8isa1  Following our related lighthouse chat a few days back I downloaded and viewed your excellent lighthouse conversion (a8isa1_brick_lighthouse_PAL.xex from May 2018), on my two CRTs and thought I'd just share them - whilst on the subject. 

Really nice conversion that! Going in my favourites folder. Detail in the building's brickwork, the clouds and also the yellows and blues is fantastic.:grin:

 

 

A8isa1's May 2018 conversion:

 

on my B&O MX4000 (800 S-video):

image.thumb.png.ecc8bff2789f4b20e4ca1a45c7b14f92.png

 

and on my 10" CRT (800XL UAV rD , S-video):

image.thumb.png.869a38468bfa70a94d7b51be4ee387f3.png

 

 

Edited by Beeblebrox
  • Like 3
Link to comment
Share on other sites

This is an image I really wanted to convert when I saw it, having seen other very good wildlife image conversions on this thread over the years. 

 

There are actually two image conversions posted here. I initially started one conversion and then some way into the run I decided to start a horizontally flipped version of the same image alongside it. (Mainly to see if the flipped version benefitted the image given the known issue sometimes with a bias in detail to the centre right of images in the conversion process).

 

Both had masks in place and some of the details in both are quite pleasing. Quite a few lines in the background, some across the frog itself in the first image, and some lack of detail on limbs/main body.

The second image is probably the better of the two, (maybe?!). Neither perfect. In hindsight I guess I should have removed the backgrounds pre-preprocessing the source images, and also maybe used some stronger dithering.

Still, worth posting em though I guess ;)).

 

I really like the greens, and how the water droplets, parts of the leaves and the upper part of the frogs have turned out. 

 

Green Frog on leaf

 

For both images:

Source file was reduced to 256 colours, RGB lines and saturation, contract and gamma tweaked/image resized to 320*240 in Irfanview, then processed in RC at 320*240 after a few more RC settings were tweaked

Settings used: Filter = bilinear; pal = atari800winplus; color dis = yuv; initial state = smart; no of solutions = 20000; Threads = 9; preprocess = yuv; brightness = 0; contrast = 4; gamma = 1.0; dither = floyd; dither strength  = 0.2; Randomness = 0.6; mask value 30

I ran Sheddy's PM line artifact removal tool on both and some errors were removed.

 

****

First up - left hand (original orientation)

 

4.3 billion evaluations (There was a problem where it didn't progress for ages so whilst the evaluations count kept ticking, the norm distance stayed the same - then it started moving again.)

69 unique colours

Filename: Beeblebrox_Frogleafsm1_11_Fixed.xex

(Altirra image with PAL high artifacting enabled)

image.thumb.png.4a7a73f6ea27547b1118d806c592cc75.png

Beeblebrox_Frogleafsm1_11_fixed.xex

 

This is the conversion running which shows the mask used.

The colours on the simulated (centre) current output image always looks a lot nicer, (more vibrant),

than the outputted XEX image or on real hardware. (eg the lower leaf is a shade of

green in RC as below but comes out a greenish blue (duck egg) colour in Altirra and real hardware):

image.thumb.png.3faea68cf0d2cfb3e46f6b1af9314c7e.png

 

****

 

Next up - Right hand flipped version which wasn't run as long as the first  - (in

some respects probably the better of the two, (fewer line errors))

 

2 billion evaluations

65 unique colours

Filename: Beeblebrox_Frogleafflippedsm2_08 fixed.xex

(Altirra image with PAL high artifacting enabled)

1380934278_Beeblebrox_Frogleafflippedsm2_08fixed.thumb.jpg.7cb4d474bc8e88f6bc274bd410c41e82.jpg

Beeblebrox_Frogleafflippedsm2_08 fixed.xex

 

Here they are on my B&O MX4000 CRT (800, S-video) Where some colours were lost in the taking of the shot on my camera phone as these look more vibrant in reality):

image.thumb.png.0337e779b52b9667e019ef4755f87f72.png   image.thumb.png.27377f6481a092a611980761fb4142b9.png   

 

And both on my 10" CRT (800XL UAV rD, S-video) where the shots are over exposed, try as I might to tweak my camera phone settings):

image.thumb.png.7d83ef5451a5753306d1c63b87944cea.png   image.thumb.png.af7574c0cead7e12f2369ce158d184dc.png 

 

Edited by Beeblebrox
  • Like 3
Link to comment
Share on other sites

  • 2 weeks later...

Here's a tip for anyone who wants to convert pictures of people, but don't want the background. There's no point RC working hard on the background when everyone wants the persons face looking decent.

 

Anyway, the tip is: Try this website, https://www.remove.bg/upload . They will remove backgrounds from your image for free. They charge for downloaded HD images, but for the size of image that we are looking at, you get it in the free preview.

 

 

  • Like 2
  • Thanks 1
Link to comment
Share on other sites

I've been doing some experiments, to see what works and what doesn't. I'm looking into cutting out backgrounds, pre-processing the input image and looking into how greyscale images can be improved (I wasn't too happy with my greyscale image for the New Years disc).

 

I've written a tool (not ready for release yet) which will:

Load a file, resize it to a resolution, typically 160x224, turn it to greyscale, reduce it to 7 colours but emphasising the pure black and pure white tones in the image, remove noise, resize it to 320x224 and then spit the file out ready for Rastaconverter.

 

I was rather happy with Forest legend Stuart Pearce with only 5 colours (I mean shades, but I'd rather call them colours)... I chopped him out of the background and making black a predominant colour on every scanline meant that Rastaconverter didn't try to hide it on his shirt.

 

snicklin-pearce-5colours-d_0.10.xeximage.png.3b3baf50ff38bfc13a555520cb28b0d9.png

 

I had tried him with 7 colours and was less impressed with how his face came out. 5 colours is almost cheating, but this is an experiment not a competition. And it needs to work out where the PMG's go on that extra colour. Considering 7 colours didn't look any better, I am happy with this. The pre-processing to the image colours helped a lot on his shirt and collar region.

 

Now here is Roy Keane in his Nottingham Forest days, with the background cut out and in colour. I've not restricted the number of colours with this and there is no pre-processing apart from the background. 75 million evaluations.

snicklin_Keane_d2.75.xeximage.png.63c6aeecfea5ebf7e3d71563e63e5ffc.png 

 

I've seen before that fleshy tones don't come out too well. I was hoping that the lack of background meant that it could use the PMG's heavily on the legs. This is still very streaky, not a pleasing result, but with some more evaluations it may get better.

 

And here he is in black and white, with pre-processing to reduce him down to 7 colours.

snicklin_Keane_BW_d0.27.xeximage.png.9aa0fb52834c5da670c91773c55663f2.png 

 

I am not too keen on how the background removal leaves that white edge around him. Though it is useful on his head so that his hair doesn't disappear into the background. His legs look more realistic in black and white.

 

My algorithm (not sure if this is how this is done in the mainstream, but it seemed to make sense was the following): If you're aiming for 7 colours, add 2 colours so that you have 9 colours. Then if all the colours are numbered 0 to 8 with 0 being black and 8 being white, all 1's are made into 0's (making the dark blacker) and all 7s are made into 8s. You then only have 7 unique colours.

 

Still plenty to learn....

 

  • Like 4
Link to comment
Share on other sites

6 hours ago, snicklin said:

Here's a tip for anyone who wants to convert pictures of people, but don't want the background. There's no point RC working hard on the background when everyone wants the persons face looking decent.

 

Anyway, the tip is: Try this website, https://www.remove.bg/upload . They will remove backgrounds from your image for free. They charge for downloaded HD images, but for the size of image that we are looking at, you get it in the free preview.

 

 

 

Tried several family pics and most worked very well. Here is one that failed miserably:

 

Andreas_Jungs.jpg.41bab83a83213bc3359b406b804c3d9d.jpg

 

Andreas_Jungs-removebg-preview.png.69ff33a79824769d6f280998b392e5e2.png

 

But I still think it's a very good tool for Rasta conversion and for TIP Animator conversion.

 

Link to comment
Share on other sites

6 hours ago, CharlieChaplin said:

 

Tried several family pics and most worked very well. Here is one that failed miserably:

 

But I still think it's a very good tool for Rasta conversion and for TIP Animator conversion.

 

 

I tried your same picture and it came back with a different result:

 

image.thumb.png.10d949f4d8487ddb83358e5353fd23db.png

Did you give it feedback? If so, the algorithm might have adapted. Thought it would better if it worked out to pick out all three of you.

Link to comment
Share on other sites

Currently running a series of space ship illustrations and this is the first. Very pleased with the colours.

 

SpaceShip (I know - imaginatively titled eh?;)

 

More to come once they finish cooking.:grin:

 

Source file was reduced to 128 colours, RGB lines and saturation, contract and gamma tweaked/image resized to 320*240 in Irfanview, then processed in RC at 320*240 after a few more RC settings were tweaked

File then fixed using Sheddys' tool to remove several line errors.

(Note - the picture taken on my B&O CRT at the bottom of this post was taken before I used Sheddys' tool to remove said pmg line errors. Just couldn't be bothered putting the fixed XEX onto the CF card again to take a new pic tbh;)).

 

2.958 billion evaluations

69 Unique colours

Filename: Beeblebrox_Spaceshipsm1_07.xex

(Altirra image with PAL high artifacting enabled)

image.thumb.png.1577f4b4b1cf2d035f90af2a8326f188.png

 

Beeblebrox_Spaceshipsm1_07_FIXED.xex

 

Settings used: Filter = bilinear, pal = altirra; color dis = yuv; initial state = smart; no of solutions = 20000; Threads = 9; preprocess = yuv; brightness = 0; contrast = 49; gamma = 1; dither = none; dither strength  = n/a; random - n/a; mask applied; mask value = 20

 

Here is is on my B&O MX4000 CRT (800, S-video):

image.thumb.png.4108010d8f8617a97b7969a2d867613c.png

Edited by Beeblebrox
  • Like 6
Link to comment
Share on other sites

Another conversion I've been running alongside the Sci-fi ones, (which are still running). Really like the colours of the building against the backdrop and the pod of killer whales in the sea. 

 

San Jan Islands, Washington, USA

 

Source file was reduced to 256 colours, RGB lines and saturation, contract and gamma tweaked/image resized to 320*200 in Irfanview, then processed in RC at 320*200 after a few more RC settings were tweaked

(For some reason when I tried to use Sheddys' other tool to centre the letterbox image onscreen it just wouldn't work.)

 

3.2 billion evaluations

60 unique colours

Filename: Beeblebrox_SanJuanIslandswhalesm1_04.xex

(Altirra image with PAL high artifacting enabled)

Beeblebrox_SanJuanIslandswhalesm1_04.thumb.jpg.ec6f560f16a55a906667741078be98fa.jpg

 

Beeblebrox_SanJuanIslandswhalesm1_04.xex

 

Settings were: height 200 letterbox, Filter = bilinear, pal = atari800winplus, color dis = yuv, initial state = smart, no of solutions = 20000, Threads = 9, preprocess = yuv, brightness = 1, contrast = 42, gamma = 0.6, dither = none, mask value = 20

 

Here is is on my B&O MX4000 CRT (800, S-video):

image.thumb.png.3585d506e44098b7b88563617312bbde.png

  • Like 4
Link to comment
Share on other sites

BTW @amarok and @drpeter- just wanted to share your recent conversions displayed on my B&O CRT/ A800 (S-video) which look great.:grin:

 

Amarok - the Toucan's colouring look amazing on a CRT, love the blue eye and claws - (and I am so glad you kept the varied green background in the conversion too btw):D:

image.thumb.png.ce0e566c6630c209cd21f4080b32da22.png

 

DrPeter - Loving the tree reflection especially. (I've wondered how a version without dithering might look too incidentally):

image.thumb.png.c1e7683dd2ee6badb11131d59798fb80.png

 

Again taking images of these CRT displays with my mobile phone camera so they look even better to the naked eye.:)

Edited by Beeblebrox
  • Like 1
  • Thanks 3
Link to comment
Share on other sites

33 minutes ago, amarok said:

@Beeblebrox, thank you very much for sharing the photo. You made my day!

 

It is the first time when I can see a result of my conversion on a real Atari, not to mention a CRT display.

Now, I have no doubt that I really need to bring my Atari 130XE back to life again.

@amarok No worries - your images are so well crafted and I know a lot of preprocessing goes into them. A lot of your conversions, (to name a few: Girl, Hummingbird, Lensball, Jaguar, and recently Winter), sit in my fav RC image folder on my Incognito's CF card as well as SIDE3. :)

 

Yup definitely get your 130XE up and running, get the best possible image out of it given stock XE displays aren't the best, then get a shielded S-video cable and if you can a decent CRT. I am always impressed with the image quality coming off my 800 on the 21" B&O MX4000. I also have a 10" Panasonic CRT which I love too - great for viewing images too.

 

Can't wait to see your next image. :lust:

 

 

Edited by Beeblebrox
Link to comment
Share on other sites

I've been away from this thread for a while guys, just catching up as I'm ready to post something again. Lot's of great conversions above! I'll take time to download and check the posts out after I finish this post. I see a new slide show version is up there too.

 

This is a re-visit to a Steampunk image I did before, someone mentioned earlier in a post it was their favorite, at least of my Steampunk conversions. This is another horizontally oriented image, to get the entire subject matter in with maximum detail still, using the entire screen instead of letter-boxing and losing so much detail in already low resolution. The first conversion was heavily cropped to just the head and shoulders region vertically oriented. This is just a first attempt I let run to about 500 million evaluations, just to get back into the swing of things. But it turned out pretty good.

 

Pretty Deadly II. 47 colors.

 

 

PrettyDeadlyII.png

GS_PrettyDeadlyIIa.xex

Edited by Gunstar
  • Like 6
Link to comment
Share on other sites

My original intent, after the first few vertically oriented images I converted to horizontally oriented images, I was going to include two or more images in each posting. One cropped more as a vertically oriented "close up" and a lesser cropped full image of vertically oriented images turned on their side horizontally.

 

But getting images of equal quality one after the other oriented and cropped differently, I haven't always been successful, in fact it's the minority like with this image. Though I intend to revisit more images that the ratio was vertically oriented and ones I cropped heavily I'll get fuller horizontal images and vice versa. So people that don't like horizontally oriented conversions can have a vertically oriented version too.

Edited by Gunstar
  • Thanks 1
Link to comment
Share on other sites

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.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...