Images generated by RastaConverter

[Gunstar]

6 hours ago, Irgendwer said:

Irgendwer_ChristmasEveSnow_PAL.xex 22.16 kB · 7 downloads

This image is absolutely incredible! As @Stephen has said about a couple of mine over the years; it looks like pixel art instead of a conversion! And there's just something about the colors...I think I have another rare favorite that I didn't do myself! And normally I don't care for conversions that use the level of dithering this one does, but it's one of the rare cases were I think it works and enhances rather than detracts, especially the color mixing the dithering is creating on this one.

Edited December 22, 2021 by Gunstar

[ilmenit]

12 hours ago, Beeblebrox said:

@_The Doctor__  Indeed. 

 

@ilmenit  I wonder if you might be aware of this phenomenon? (As detailed in my post above re the partially resetting evaluation counter?)

I think you guys are totally crazy with how long you let RC running ? Seems that you hit the integer overflow as I didn't expect anyone will keep RC working for that long. 

[Beeblebrox]

@ilmenit  Thanks for the explanation. Yup, you have created an army of obsessive monsters in giving us this amazing tool! 

[Beeblebrox]

Speaking of obsessive, I recently dug out my beloved ol Atari Jaguar console and the game Alien vs Predator, (still scares the cr@p outta me playing this title on the Jag!). It reminded me of this image from the box art and game............ so I had a go at converting it.

Really pleased with the result. I stopped this one earlier as I would have ideally run it longer, but had other images to finish. So a few line errors, etc, but IMHO it doesn't detract from the overall image too much. It got to a Norm dist of 14. I am sure a little more detail would have come through had I left it as is my experience with some of the other images I have run/am currently cooking.

 

Incidentally this was without any mask and before I'd switched back to using Altirra pallette. I think the colours have benefitted from the Atari800winplus pallette though, as the source image is very garish and vibrant anyway. If I revisited this one I'd use a mask next time for the characters for sure.

 

Source file was reduced to 256 colours then resized to 160x240 and then processed in RC at 320*240. RGB lines, contrast, gamma and saturation generally ramped up in Irfanview, then some more tweaking in RC's GUI settings, till I was happy with the destination image.

 

3+ Billion evaluations

70 Unique colours

Filename: Beeblebrox_AVPsm1_16.xex

Beeblebrox_AVPsm1_16.xex

 

Settings were:

Filter = box
pal = atari800winplus
color dis = yuv
initial state = smart
no of solutions = 20000
Threads = 9
preprocess = yuv
brightness = 1
contrast = 46
gamma = 1.2
dither = floyd
dither strength  = 1.0

 

This it was it looks like running off my 800 on my B&O MX4000 crt, (again mobile phone camera doesn't do it justice):

 

Edited December 22, 2021 by Beeblebrox

[Irgendwer]

2 hours ago, ilmenit said:

I think you guys are totally crazy with how long you let RC running ? Seems that you hit the integer overflow as I didn't expect anyone will keep RC working for that long. 

While you're browsing this thread:

 

I know that you are aware of the glitches RC produces - inspect the real image here:

 

You said, fixing this in the optimized version (from @phaeron ) is very unlikely to happen. But what about fixing it in the not optimized one?

I'm sure volunteers (and so maybe phaeron again) will go another round to incorporate your changes.

It's really a pity, that all the calculation time spend has a quite big chance to produce a picture with little defects.

 

I think this thread is a testimonial how popular your work is. So if I could wish a Christmas present from you, you know what it would be....

 

Edited December 22, 2021 by Irgendwer

[drpeter]

On 12/21/2021 at 9:17 AM, Beeblebrox said:

I just wondered if anyone else has experienced it?

Yes

[drpeter]

32 minutes ago, Irgendwer said:

While you're browsing this thread:

 

I know that you are aware of the glitches RC produces - inspect the real image here:

 

You said, fixing this in the optimized version (from @phaeron ) is very unlikely to happen. But what about fixing it in the not optimized one?

I'm sure volunteers (and so maybe phaeron again) will go another round to incorporate your changes.

It's really a pity, that all the calculation time spend has a quite big chance to produce a picture with little defects.

 

I think this thread is a testimonial how popular your work is. So if I could wish a Christmas present from you, you know what it would be....

 

This image has an unusual number of artefacts- 13 in total, including those highlighted here and 10 less visually obvious ones.

 

Using the fabulous tools made available by @Sheddy, took less than 15 mins to fix 'em all by hand:

 

 

Irgendwer_ChristmasEveSnow_PAL_fixed_drpeter.xex

 

Using @Sheddy's automated tool also removed all of them bar 1, including all the visible ones, in less than 1 minute! ?

Edited December 22, 2021 by drpeter

[Beeblebrox]

1 hour ago, Irgendwer said:

While you're browsing this thread:

 

I know that you are aware of the glitches RC produces - inspect the real image here:

 

You said, fixing this in the optimized version (from @phaeron ) is very unlikely to happen. But what about fixing it in the not optimized one?

I'm sure volunteers (and so maybe phaeron again) will go another round to incorporate your changes.

It's really a pity, that all the calculation time spend has a quite big chance to produce a picture with little defects.

 

I think this thread is a testimonial how popular your work is. So if I could wish a Christmas present from you, you know what it would be....

 

Just to quote DrPeter from back in May regarding any revisit from either Ilmenit and Phaeron:

https://atariage.com/forums/topic/200118-images-generated-by-rastaconverter/?do=findComment&comment=4827911

 

.... but great re Sheddy's tool(s) - I'll check them out as automatic removal of a large percentage of these would be great. 

 

[Irgendwer]

50 minutes ago, drpeter said:

Using @Sheddy's automated tool also removed all of them bar 1, including all the visible ones, in less than 1 minute! ?

Thanks for the fix and of course @Sheddy for his tool!

 

Nevertheless I would appreciate a revisit of RC anyway (like most of you), as a different/updated view on the timing could produce other/better images?

 

12 minutes ago, Beeblebrox said:

Just to quote DrPeter from back in May regarding any revisit from either Ilmenit and Phaeron:

Yes, it's an elaborated art to kindly remind someone of something contrary to jangle someones nerves.

I thought a nice but imperfect conversion would be a good occasion...

 

[Beeblebrox]

One of the new conversions I've recently ran trialing the use of masks going forward.

Galadriel from Lord of the Rings.

 

Source image reduced to 256 colours resized to 320*240 and reduced to 256 colours. RGB lines, contrast, gamma and saturation generally tweaked in Irfanview, then some more tweaking in RC's GUI settings, till I was happy with the destination image. I didn't quite make the mask big enough to cover a small part of the neck under the jawline so there is a notable brown shading which might have had more graduated detail there had I done so. Skin tones a little hard to nail based on the source image but no too bad. (Source image was a picture of a LOTR statuette). If I did it again I'd likley pre-edit the source image to bring down the brightness and contrast of the face and neck tones independantly of the rest of the image.

 

1.8+ Billion evaluations

35 Unique colours

Filename: Beeblebrox_Galadrielsm1_12.xex

Beeblebrox_Galadrielsm1_12.xex

 

Settings were:

Filter = bicubic
pal = altirra
color dis = yuv
initial state = smart
no of solutions = 20000
Threads = 9
preprocess = yuv
brightness = -1
contrast = 35
gamma = 0.9
dither = floyd
dither strength  = 1.0
Mask value = 10

 

This it was it looks like running off my 800 on my B&O MX4000 crt, (again as mentioned before mobile phone camera doesn't do it justice - yellow seems to be dominating more than in reality):

[Beeblebrox]

46 minutes ago, Irgendwer said:

 

Yes, it's an elaborated art to kindly remind someone of something contrary to jangle someones nerves.

I thought a nice but imperfect conversion would be a good occasion...

 

@Irgendwer  I totally agree the RC would be ever more amazing if it were revisited - I would happily welcome it as it is one of the most amazing tools IMHO and grants us the ability to realise some amazing graphics on our lil A8. RC is what hooked me back onto the A8 scene earlier this year. As you can no doubt tell I am somewhat obsessed with it heh heh

 

I terms of refencing the previous post by DrPeter regarding the stance of both Ilmenit and Phaeron on RC, I was merely just avoiding repeating it myself. I often signpost other's posts to be helpful. I meant no offence, (if indeed offence was taken... I am not quite sure if it was ). So, no offence was intended and I certainly wasn't intending to jangle anyones nerves...... that's not my style 

Edited December 22, 2021 by Beeblebrox

[ilmenit]

2 hours ago, Irgendwer said:

I think this thread is a testimonial how popular your work is. So if I could wish a Christmas present from you, you know what it would be....

 

One day I may return to this project, however it would require significant investment of time that I do not have currently - if fixing of it was easy everyone would do it till now. A while ago also @fox wrote on AtariOnline that he may try to create a variant of RC for the GPU compiler, to significantly speed up calculations.

[Beeblebrox]

Another new image using a mask.

Vintage diving suit helmet. (I now always think of the Bioshock games when I see these!) 

 

In Irfanview source image resized to 320*240, colours decreased to 256. RGB lines and contrast, gamma, saturation tweaked, then in RC GUI settings tweaked further.

Mask applied to the whole outline of the diving helment, although in hindsight given this is an image on a black background I am not sure a mask was needed at all. Ah well - live and learn!

 

Conversion was stopped fairly early on as other images to convert. Thought worth posting up as quite a nice conversion, (I quite like the little front glass window).

 

589 million evalation

52 unique colours

Filename: Beeblebrox_Divershemlmetsm2_04.xex

Beeblebrox_Divershemlmetsm2_04.xex

 

Settings were:
height 240
Filter = bicubic
pal = altirra
color dis = yuv
initial state = smart
no of solutions = 15000
Threads = 9
preprocess = yuv
brightness = 2
contrast = 50
gamma = 0.9
dither = floyd
dither strength  =1.4
Mask value = 10

 

... and on my B&O MX4000 CRT (Atari 800):

 

 

Edited December 22, 2021 by Beeblebrox

[Wrathchild]

NTSC version... needs some repairs, e.g. remove lighter blue in sky, but not too bad

 

Wrathchild_PolarExpress_NTSC.xex

[drpeter]

3 hours ago, Beeblebrox said:

given this is an image on a black background I am not sure a mask was needed at all

There is definitely an art to the judicious choice of when to use a mask and where.

 

It's important to keep in mind that what the mask is essentially doing is to nudge the Rastaconverter algorithm into accepting solutions that are more accurate where the mask is strongest at the expense of accepting less accuracy in areas nearby, especially those nearby horizontally.  A mask is unlikely to magically improve overall accuracy or detail across the board.

 

I will sometimes use a mask from the start, where past experience indicates that one important element of the picture will almost certainly suffer for the lack of it- often for central facial features, for example.  Girl With Headphones was a good example of this- I went from the start for accurate detail on the face at the expense of the hair.

 

With many images I will do a couple of Rastaconverter runs first to see if the conversions have a tendency to repeatedly produce unresolvable visually jarring inaccuracies in an important element of the image before deciding to mask it.  Tutankhamun was a good example of this- I ended up applying a moderate mask to the main facial features and beard, and to an area of peripheral background lower right that otherwise tended to accumulate dark horizontal streaking 'bleeding out' from the edge of the coffin.

 

One thing I have learned is to overlap the mask slightly with the area adjacent to that you are interested in, otherwise the algorithm will tend to generate jarring streaks at the interface. For example, in Girl With Headphones the mask is extended slightly over the background adjacent to the edges of the face- without that, streaks of colour from the face tended to extend between the face and the right border of the image.

 

 

[Gunstar]

9 hours ago, Beeblebrox said:

One of the new conversions I've recently ran trialing the use of masks going forward.

Galadriel from Lord of the Rings.

 

Source image reduced to 256 colours resized to 320*240 and reduced to 256 colours. RGB lines, contrast, gamma and saturation generally tweaked in Irfanview, then some more tweaking in RC's GUI settings, till I was happy with the destination image. I didn't quite make the mask big enough to cover a small part of the neck under the jawline so there is a notable brown shading which might have had more graduated detail there had I done so. Skin tones a little hard to nail based on the source image but no too bad. (Source image was a picture of a LOTR statuette). If I did it again I'd likley pre-edit the source image to bring down the brightness and contrast of the face and neck tones independantly of the rest of the image.

 

1.8+ Billion evaluations

35 Unique colours

Filename: Beeblebrox_Galadrielsm1_12.xex

Beeblebrox_Galadrielsm1_12.xex 22.1 kB · 6 downloads

 

Settings were:

Filter = bicubic
pal = altirra
color dis = yuv
initial state = smart
no of solutions = 20000
Threads = 9
preprocess = yuv
brightness = -1
contrast = 35
gamma = 0.9
dither = floyd
dither strength  = 1.0
Mask value = 10

 

This it was it looks like running off my 800 on my B&O MX4000 crt, (again as mentioned before mobile phone camera doesn't do it justice - yellow seems to be dominating more than in reality):

Something I've never bothered doing was trying anything but the default box filter.

 

Is there ever a reason not to use the same color distance in pre-processing than you intend to convert it? I always thought having to choose for both a waste of time and just choose one color distance that the both will use, just like with the palette. Just an observation as I looked down your settings list and I see you have them both set the same as well. Though if there is a reason to use a different color distance in pre-processing than during conversion it's beyond me and I'd like to know.

Edited December 22, 2021 by Gunstar

[Beeblebrox]

5 minutes ago, Gunstar said:

Something I've never bothered doing was trying anything but the default box filter.

 

Is there ever a reason not to use the same color distance in pre-processing than you intend to convert it? I always thought having to choose for both a waste of time and just choose one color distance that the both will use, just like with the palette. Just an observation as I looked down your settings list and I see you have them both set the same as well.

@Gunstar  I tend to flip between Box, Bicubic and Bilinear as I noticed a subtle difference in the previewing.  I use to use Box all the time but recently I tend to use bicubic and bilinear in equal measure.

 

Regarding the pre-processing and colour distance values - I suppose I got into the habit of record all the settings down for each conversion, hence why preprocessing is recorded. I'll can confirm I never (intentionally) chose a different value here. So at the moment I am using YUV for both in my current recent and running conversions.

 

Whilst the colour distance value is the only one used in the actual final conversion, pre-processing always defaults to Ciede when you load up an instance of RC GUI for the first time. So I have to change pre-processing to YUV if that is my intended colour distance value as because I use the pre-processing preview anywhere between 10-50 times whilst tweaking settings in RC before I then hit convert. So I always match them up so in this fine tuning stage I can see the YUV preview for an intended YUV based conversion each time, rather than a Ceide based preview. The preview and tweak stage IMHO is the most important part of the pre-processing exercise after tweaking the source image in something like Irfanview, (which incidentally has been by fav free image tool for well over 15 years).

 

BTW in the GUI if you preview in Ceide the preview calculation image takes around 10 seconds to appear, (and even longer with dither engaged, especially with a knoll and ciede combo), whereas previewing with YUV takes 3-5 seconds. 

 

BTW I've never used the original command line version of RC which I think a lot of people use. I just decided from the outset to install the GUI version as I prefer GUIs over command line and because it is so easy to tweak and preview multiple times before hitting convert.

 

[Gunstar]

49 minutes ago, Beeblebrox said:

@Gunstar  I tend to flip between Box, Bicubic and Bilinear as I noticed a subtle difference in the previewing.  I use to use Box all the time but recently I tend to use bicubic and bilinear in equal measure.

 

Regarding the pre-processing and colour distance values - I suppose I got into the habit of record all the settings down for each conversion, hence why preprocessing is recorded. I'll can confirm I never (intentionally) chose a different value here. So at the moment I am using YUV for both in my current recent and running conversions.

 

Whilst the colour distance value is the only one used in the actual final conversion, pre-processing always defaults to Ciede when you load up an instance of RC GUI for the first time. So I have to change pre-processing to YUV if that is my intended colour distance value as because I use the pre-processing preview anywhere between 10-50 times whilst tweaking settings in RC before I then hit convert. So I always match them up so in this fine tuning stage I can see the YUV preview for an intended YUV based conversion each time, rather than a Ceide based preview. The preview and tweak stage IMHO is the most important part of the pre-processing exercise after tweaking the source image in something like Irfanview, (which incidentally has been by fav free image tool for well over 15 years).

 

BTW in the GUI if you preview in Ceide the preview calculation image takes around 10 seconds to appear, (and even longer with dither engaged, especially with a knoll and ciede combo), whereas previewing with YUV takes 3-5 seconds. 

 

BTW I've never used the original command line version of RC which I think a lot of people use. I just decided from the outset to install the GUI version as I prefer GUIs over command line and because it is so easy to tweak and preview multiple times before hitting convert.

 

Uhh, yep. I know all of that and all of what you just described about color distances there is exactly what I meant by my comments. And yes, I make sure I use the same color distance on both too, and what they are used for, in my understanding, you don't want to preview in a different color distance as it won't be giving you a "preview" of the image in the color distance you use in conversion, if it is different. So why why not have just the one option to choose the color distance for both, just like happens with choosing a palette, or filtering? I understand if it always defaults to Ceide, whether it has too for some reason or just does, but why two separate places to set color distance so you have to make sure they both match up? IS there some possible reason?

 

My earlier comments were meant as a half joke, but if there is a reason to have to choose color distance twice I would like to know. But that's just sometimes my humour is a bit too dry and people don't catch it as that and think I am serious.

 

As to the command line, no, I've always used the GUI too. I'll use command lines if that's all there is and do for some things, and of course I use command lines with SpartaDOS X all the time, but on PC's I'll use a GUI if the app has one. I used to use command lines a lot on MS-DOS back in the first decade of the 21st century for burning 90's CD console software onto CD-R's and for using an original Atari developer's flashcart for the Jaguar and for the early skunkboards.

 

But of course these days, even if I do use my Skunkboard, I use Jiffy for the Jaguar with it's GUI. But mostly I have SD or CF card drives for my computers and consoles now, like the Jaguar Game Drive and 7800 Dragonfly, etc., etc. and don't have to burn CDR's for my 90's consoles that had CD anymore for that reason and there hasn't been anything since that I use the doesn't have a GUI front to it, and I'm just in that habit now instead of the command line habit, except for the Atari 8-bit and SDX. And soon the Coco 2 as well once I get my upgrading it wrapped up and actually start using it.

Edited December 22, 2021 by Gunstar

[Beeblebrox]

10 minutes ago, Gunstar said:

Uhh, yep. I know all of that and all of what you just described about color distances there is exactly what I meant by my comments. And yes, I make sure I use the same color distance on both too, and what they are used for, in my understanding, you don't want to preview in a different color distance as it won't be giving you a "preview" of the image in the color distance you use in conversion, if it is different. So why why not have just the one option to choose the color distance for both, just like happens with choosing a palette, or filtering? I understand if it always defaults to Ceide, whether it has too for some reason or just does, but why two separate places to set color distance so you have to make sure they both match up? IS there some possible reason?

 

My earlier comments were meant as a half joke, but if there is a reason to have to choose color distance twice I would like to know. But that's just sometimes my humour is a bit too dry and people don't catch it as that and think I am serious.

Ah no worries. I probably read you post too quickly.   (Plus can a little tricky to gauge some things in forum posts, humour being one of them ). Been a long day ... cleaning the house in anticipation of us hosting Christmas this year... I am probably high on cleaning chemicals!

 

Also sometimes I am not entirely sure where if another RC user is using the command line version of RC, whether their experience is very different from the GUI version. 

 

Anyways, all this aside - yeah, I agree, it makes sense to only have the one setting (colour distance), where the preview just defaults to the same - so not sure why there are the two available. Perhaps Ilmenit threw it in there just to confuse us. 

[Gunstar]

41 minutes ago, Beeblebrox said:

Ah no worries. I probably read you post too quickly.   (Plus can a little tricky to gauge some things in forum posts, humour being one of them ). Been a long day ... cleaning the house in anticipation of us hosting Christmas this year... I am probably high on cleaning chemicals!

 

Also sometimes I am not entirely sure where if another RC user is using the command line version of RC, whether their experience is very different from the GUI version. 

 

Anyways, all this aside - yeah, I agree, it makes sense to only have the one setting (colour distance), where the preview just defaults to the same - so not sure why there are the two available. Perhaps Ilmenit threw it in there just to confuse us. 

Yeah, text often makes it harder to tell what someone really means or mood. But it's probably just you read through too fast as English seem to pick up on my sense of dry humor easily. I first discovered this in college when I always hit it off so well with the English exchange students. I could say something to American's and the result was more often rolling eyes or shake of the head or dumb-founded expression, and say the same thing to an Englishman and straight away get the proper bellowing laughter I expected, or at least hoped for.?

Edited December 22, 2021 by Gunstar

[amarok]

Wishing you a Merry Christmas and Happy New Year! May all the happiness of this world find its way to all of you!

 

 

amarok_winter.xex

[Beeblebrox]

I am sure this is the case for anyone who has gotten obsessed with Rastaconverter and seen an image they really wanted to convert for the A8.

 

I've tried no less than 6 separate times over the course of this year to convert this image.

Either due to wrong settings, inexperience, or the other week a drunken mishap post work Christmas meal which saw me wipe the entire conversion and lose days of work, this image has aluded me through the Spring, Summer and Winter. It was I grant you incredibly ambitious/foolhardy to try, given the multicolour nature of the source image. However on this 6th attempt I think I've gotten as close to it as I have ever done. I am so pleased with how this has turned out.

 

It might have lots of artifacts, especially horizontally across the legs which crept in towards the end, and others might not rate it for these reasons among others, ....(I totally acknowledge it's far from perfect); but I personally love it despite it's flaws, and after 9.5 billion evaluations and a whopping 113 unique colours personally I think seeing this on my A8s is one of the most rewarding things IMHO. 

 

Incidentally I could have converted the image without the cityscape background and instead have a black/blank background. I've done black/blank backgrounds before for a lot of my convesions and it works well. However for me the cityscape backdrop is essential for this image, not least because it visually gives the lightsources and tones on the main character a context. The golden colours also offset the main character's colours IMHO.

 

Thanks to @drpeter for initially looking over this for any possible line error fixes, which it turned out wasn't possible. 

 

Incidentally if I ever attempted a 7th conversion I'd make sure the Mask covered both legs as well.

 

Source image pre-processing:

Image reduced to 256 colours then resized to 160*240 then in RC processed at 320*240

RGB lines, Contrast, gamma and saturation tweaked in Irfanview then further tweaks in RC settings.

 

9.5 billions evaluations

113 unique colours, (highest ever in any of my conversions)

Filename: Beeblebrox_Multicolouredspidermansm6_31.xex

Beeblebrox_Multicolouredspidermansm6_31.xex

 

Settings were:

Filter = bicubic
pal = altirra
color dis = yuv
initial state = smart
no of solutions = 20000
Threads = 9
preprocess = yuv
brightness = 5
contrast = 49
gamma = 1.2
dither = jarvis
dither strength  = 1.2
Mask value = 10

 

Zoomable version (screen dump from my Window's laptop running Altirra - with Pal artifacting setting enabled):

 

... and here is it displayed on my B&O MX4000 on my Atari 800, (left), with Altirra on the right, (again with Pal artifacting setting enabled):

 

(Again as ever my mobile phone camera loses some of the colours that are actually present on the CRT output - it looks more vibrant that this in reality)

 

Here is the live conversion, showing the tweaked source image, current output and destination image, including how the mask was applied:

 

...and here is the original untouched full resolution source image before it was pre-processed/tweaked, etc, (you can probably see why I was so intrigued by it ):

 

 

 

Edited December 23, 2021 by Beeblebrox

[Beeblebrox]

A version of the Doom title screen, (from a Sega version of the game I think) - very pleased with the colours and the definition overall, especially that of the DOOM guy character, (thanks to the mask used). You can make out his eyes and the reflection on the visor.  Damn I love using masks!!

 

This is the 4th attempt this year atgetting this one close to presentable. Great to see this displayed on my A8s - I love anything to do with DOOM .

 

Pity about the 4 horizontal lines and a few glitches but ya can't have everything. Still they are dark so you don't notice them that much.

 

1.530 billion evaluations

71 Unique colours

Filename: Beeblebrox_Doomtitleblackbgsm3_07.xex

Beeblebrox_Doomtitleblackbgsm3_07.xex

 

Settings were:

Filter = bicubic
pal = altirra
color dis = yuv
initial state = smart
no of solutions = 20000
Threads = 9
preprocess = yuv
brightness = 0
contrast = 29
gamma = 1
dither = floyd
dither strength  =0.5
mask value 10

 

This was an earlier pic taken from a slightly earlier save (yesterday), in the conversion process on my B&O MX4000 CRT off my 800:

[+JAC!]

On 12/20/2021 at 1:58 AM, drpeter said:

Tutankhamun

 

drpeter_Tutankhamun.xex 22.02 kB · 20 downloads

 

On 12/22/2021 at 12:08 AM, Irgendwer said:

Irgendwer_ChristmasEveSnow_PAL.xex 22.16 kB · 20 downloads

 

On 12/23/2021 at 12:25 AM, amarok said:

Wishing you a Merry Christmas and Happy New Year! May all the happiness of this world find its way to all of you!

 

 

amarok_winter.xex 22.15 kB · 18 downloads

Wow, feels like Christmas - all of these are absolutely incredible. Thanks you guys for these masterpieces!

[Irgendwer]

Irgendwer_Bled_PAL.xex