AI frames with Red Dwarf auras

[gerrydelasel]

Sirs, I have a little ongoing project, which is to create high-res images of Red Dwarf using a combination of Ai and photoshop. So I’ve created this thread to avoid tarnishing the other threads. Please, don’t hit me.

Topic

[gerrydelasel]

[Jonathan Capps]

Thanks for keeping the AIness contained. The pug’s honour has been preserved.

[gerrydelasel]

[Quinn: Clochebusters World Champion]

This image but with the those truck nuts thing in place of the asteroid 

[Danny Stephenson]

I would have killed for empty set pictures like these when I made my Smegopoly board!

I had to do it the old fashioned way and take screenshots when the cast were in different parts of the set and then just cobble together the empty sets from those like a jigsaw.

I had a lot of time on my hands *counts on fingers* *runs out of fingers* 19 years ago. Fuck me, I’m old.

[Ian Symes]

Yeah, but at least in yours, the light above Lister’s bunk didn’t read “IFC TOGA ANRONT []”

[Moonlight]

[Moonlight]

Yeah, but at least in yours, the light above Lister’s bunk didn’t read “IFC TOGA ANRONT []”

Gonna be slightly cunty and point out how incredibly easy this would be to fix with Photoshop or GIMP or whatever.

[Ridley]

Fuck me, I’m old.

Not here, it’s too sandy. 

[Moonlight]

That exchange is like 90% of my interactions with other people.

[gerrydelasel]

[gerrydelasel]

[Unrumble]

[gerrydelasel]

[gerrydelasel]

[gerrydelasel]

[gerrydelasel]

[Dave]

Norman’s had some work done.

[Podey]

[gerrydelasel]

Considering he’s been filmed on video tape, then played on a small CRT, in a dark corner, while moving, I’d say it’s a pretty good likeness.

[Dave]

But if you’re piecing these together yourself, couldn’t you just chuck in a better-quality close-up of Holly on the monitor from another shot, rather than relying only on the version from a wide angle?

Don’t get me wrong, some of these are nicely sharp looking versions of shots that don’t actually exist in this quality, but that makes easily-fixed AI artifacts like Holly and the bunk-light text stand out even more.

[gerrydelasel]

Yes. It’s all a work in progress. If the worst criticism is one tiny bit of text in a whole image full of other text, that you can only make out by zooming in, I think I’m doing pretty well.

[gerrydelasel]

[gerrydelasel]

[gerrydelasel]

[Ian Symes]

Sorry, but this definitely doesn’t look like Norman.

[gerrydelasel]

[Dave]

Sorry, but this definitely doesn’t look like Norman.

[Unrumble]

[gerrydelasel]

[gerrydelasel]

[Frank Smeghammer]

This is Remastered all over again

[gerrydelasel]

[Flap Jack]

The frame rate was a little better in Remastered.

[gerrydelasel]

This one is a real pain because of the wacky geometry of the room. Best I can do at the moment.

[gerrydelasel]

[Podey]

 If the worst criticism is one tiny bit of text in a whole image full of other text, that you can only make out by zooming in, I think I’m doing pretty well.

I think a lot of us have far worst criticisms when it comes to the use of generative AI but we’re being polite.

[gerrydelasel]

[Unrumble]

[gerrydelasel]

I’m quite pleased with this one. I nicked a photo of an original starbug model from the propstore, asked AI to adjust the lighting, then I stuck it on a modified BBC starry background and pulled down the saturation. I think it’s damn close to an early season model flyby. With enough of these I should be able to create new model footage.

[gerrydelasel]

Marooned is my favourite episode

[gerrydelasel]

[gerrydelasel]

[gerrydelasel]

Creating an image of the Starbug cockpit is incredibly difficult. The footage is really low res, dark, always full of people, and of course AI stuggles with the concept of ejector seats. It’s taken me a long time to create this. I’m not entirely happy with it, but it’s not terrible.

[Unrumble]

While model/set shots pass muster at a glance, this highlights the issue with a lot of AI-generated faces, almost immediately you notice a too-smooth-with-a-capital-smoo waxiness.

[gerrydelasel]

It does tend to do that, yes, but black skin does have a sheen to it, and with makeup under hot studio lights I don’t think that is a good example to criticise. This on the other hand…

[FileTrekker]

I mean, I have to say that you seem really proud of yourself, but, all you’re doing here is giving an AI agent some prompts and letting it do 99% of the work. If there’s any photoshop skills here, they’re incredibly basic like tweaking some saturation and contrast, because it would have taken anyone who knew how to use Photoshop a couple minutes at most to fix the text on the bunkroom light, Holly on the monitor, and the multitude of other AI generated crimes in these images.

Just be wary of the credit you’re taking here. It’s a fun experement to see what it may look like in pseudo-HD but, all of these images are just really strange and problematic. I mean, this waxwork of Mark Williams, oh, and why is Officer Crabtree having a conversation with Vin Diesel?

[gerrydelasel]

Dude, I did photoshop the text in the smegging bunk room. Not to mention all the other defects, objects, and details that had to be photoshopped in or out. You can’t get that image just from an AI prompt. The fact you can’t see the photoshopping is because its practically seamless. I just missed that one bit of text for fucks sake!

[FileTrekker]

Ah, that must be why Zannta Broccli’s swimming certificates look like they’re written in German, and there’s a copy of Bwynino67, everyone’s favourite magazine, on the table there.

Brb, off to watch the Lauhnnn Jets.

[gerrydelasel]

Once the text is so pixelated that it’s not readable I don’t see the point of editing it, how do you know what it’s ‘supposed’ to say if it’s just blur?

[Author]

Replies