ChaoticNeutralCzech

joined 2 years ago
[–] [email protected] 1 points 1 week ago

You're right. Later in the video, this shot with the same fake film effect appears and that's indubitably AI (look at bottom right):

The video narration implied this is footage from a rare or unfinished film, though.

[–] [email protected] 2 points 1 week ago* (last edited 1 week ago) (1 children)

Which spoiler works in Thunder?

Lemmy syntax

    
spoiler Lemmy syntax <content> :::

:::

Reddit syntax: >!>!<content>!<!<

[–] [email protected] 5 points 1 week ago (2 children)

Are you implying it's from a stock footage site that used AI? They would definitely get their previews indexed on search engines. Alternatively, it's been generated on request, which would make it impossible to find.

[–] [email protected] 1 points 1 week ago (4 children)

I put them in a spoiler. Compliant viewers should hide them by default.

[–] [email protected] 5 points 1 week ago (1 children)

The film artifacts are quite unusual (the vertical lines span exactly one frame, one of the lighter spots stays on pretty much the same spot between frames 1 and 2) but I noticed no other red flags. In fact, the hair is very convincing. The eyes seem to reflect different things but her right one is somewhat in the shade and the light source reflection could be different because it's close to her face.

 

Found by @[email protected]: it originates here: Dune by Alejandro Jodorowsky - Teaser Trailer (1976)


Source: used as B-roll in the intro of this video: https://www.youtube.com/watch?v=f8AJk2Sns_k&t=3

Here are individual frames but image search (SauceNAO, Google Lens, IQDB, Yandex) has not been helpful.

Frames




























Transcript: Close-up shot of a woman's face with a neutral expression, short brown '80s hair, lipstick, thick sharp eyeliner and glowing aqua irises. Widescreen with a higher-than-usual amount of film artifacts.

[–] [email protected] 1 points 1 week ago

Maybe I'll find others but it's over for now. Have I been spamming?

[–] [email protected] 2 points 2 months ago

commuting

TIL gazebos can live and work in different places

 

Artist: Onion-Oni aka TenTh from Random-tan Studio
Original post: Electric girls on Tapas (warning: JS-heavy site)

Upscaled by waifu2x (model: upconv_7_anime_style_art_rgb). Original

This is the last one in the series. Bye!

 

Artist: Onion-Oni aka TenTh from Random-tan Studio
Original post: Electric girls on Tapas (warning: JS-heavy site)

Upscaled by waifu2x (model: upconv_7_anime_style_art_rgb). Original
Unlike photos, upscaling digital art with a well-trained algorithm will likely have little to no undesirable effect. Why? Well, the drawing originated as a series of brush strokes, fill areas, gradients etc. which could be represented in a vector format but are instead rendered on a pixel canvas. As long as no feature is smaller than 2 pixels, the Nyquist-Shannon sampling theorem effectively says that the original vector image can therefore be reconstructed losslessly. (This is not a fully accurate explanation, in practice algorithms need more pixels to make a good guess, especially if compression artifacts are present.) Suppose I gave you a low-res image of the flag of South Korea 🇰🇷 and asked you to manually upscale it for printing. Knowing that the flag has no small features so there is no need to guess for detail (this assumption does not hold for photos), you could redraw it with vector shapes that use the same colors and recreate every stroke and arc in the image, and then render them at an arbitrarily high resolution. AI upscalers trained on drawings somewhat imitate this process - not adding detail, just trying to represent the original with more pixels so that it loooks sharp on an HD screen. However, the original images are so low-res that artifacts are basically inevitable, which is why a link to the original is provided.

 

Artist: Onion-Oni aka TenTh from Random-tan Studio
Original post: Electric girls on Tapas (warning: JS-heavy site)

Upscaled by waifu2x (model: upconv_7_anime_style_art_rgb). Original
Unlike photos, upscaling digital art with a well-trained algorithm will likely have little to no undesirable effect. Why? Well, the drawing originated as a series of brush strokes, fill areas, gradients etc. which could be represented in a vector format but are instead rendered on a pixel canvas. As long as no feature is smaller than 2 pixels, the Nyquist-Shannon sampling theorem effectively says that the original vector image can therefore be reconstructed losslessly. (This is not a fully accurate explanation, in practice algorithms need more pixels to make a good guess, especially if compression artifacts are present.) Suppose I gave you a low-res image of the flag of South Korea 🇰🇷 and asked you to manually upscale it for printing. Knowing that the flag has no small features so there is no need to guess for detail (this assumption does not hold for photos), you could redraw it with vector shapes that use the same colors and recreate every stroke and arc in the image, and then render them at an arbitrarily high resolution. AI upscalers trained on drawings somewhat imitate this process - not adding detail, just trying to represent the original with more pixels so that it loooks sharp on an HD screen. However, the original images are so low-res that artifacts are basically inevitable, which is why a link to the original is provided.

[–] [email protected] 9 points 2 months ago (1 children)
 

Artist: Onion-Oni aka TenTh from Random-tan Studio
Original post: Gazebo on Tapas (warning: JS-heavy site)

Upscaled by waifu2x (model: upconv_7_anime_style_art_rgb). Original
Unlike photos, upscaling digital art with a well-trained algorithm will likely have little to no undesirable effect. Why? Well, the drawing originated as a series of brush strokes, fill areas, gradients etc. which could be represented in a vector format but are instead rendered on a pixel canvas. As long as no feature is smaller than 2 pixels, the Nyquist-Shannon sampling theorem effectively says that the original vector image can therefore be reconstructed losslessly. (This is not a fully accurate explanation, in practice algorithms need more pixels to make a good guess, especially if compression artifacts are present.) Suppose I gave you a low-res image of the flag of South Korea 🇰🇷 and asked you to manually upscale it for printing. Knowing that the flag has no small features so there is no need to guess for detail (this assumption does not hold for photos), you could redraw it with vector shapes that use the same colors and recreate every stroke and arc in the image, and then render them at an arbitrarily high resolution. AI upscalers trained on drawings somewhat imitate this process - not adding detail, just trying to represent the original with more pixels so that it loooks sharp on an HD screen. However, the original images are so low-res that artifacts are basically inevitable, which is why a link to the original is provided.

 

Artist: Onion-Oni aka TenTh from Random-tan Studio
Original post: The Infinity Gauntlet on Tapas (warning: JS-heavy site)

Upscaled by waifu2x (model: upconv_7_anime_style_art_rgb). Original
Unlike photos, upscaling digital art with a well-trained algorithm will likely have little to no undesirable effect. Why? Well, the drawing originated as a series of brush strokes, fill areas, gradients etc. which could be represented in a vector format but are instead rendered on a pixel canvas. As long as no feature is smaller than 2 pixels, the Nyquist-Shannon sampling theorem effectively says that the original vector image can therefore be reconstructed losslessly. (This is not a fully accurate explanation, in practice algorithms need more pixels to make a good guess, especially if compression artifacts are present.) Suppose I gave you a low-res image of the flag of South Korea 🇰🇷 and asked you to manually upscale it for printing. Knowing that the flag has no small features so there is no need to guess for detail (this assumption does not hold for photos), you could redraw it with vector shapes that use the same colors and recreate every stroke and arc in the image, and then render them at an arbitrarily high resolution. AI upscalers trained on drawings somewhat imitate this process - not adding detail, just trying to represent the original with more pixels so that it loooks sharp on an HD screen. However, the original images are so low-res that artifacts are basically inevitable, which is why a link to the original is provided.

15
submitted 2 months ago* (last edited 2 months ago) by [email protected] to c/[email protected]
 

Artist: Onion-Oni aka TenTh from Random-tan Studio
Original post: Frostpunk Automaton on Tapas (warning: JS-heavy site)

Upscaled by waifu2x (model: upconv_7_anime_style_art_rgb). Original

See also: Land Dreadnought

19
submitted 3 months ago* (last edited 3 months ago) by [email protected] to c/[email protected]
 

Artist: Onion-Oni aka TenTh from Random-tan Studio
Original post: Crabsquid on Tapas (warning: JS-heavy site)

Upscaled by waifu2x (model: upconv_7_anime_style_art_rgb). Original

See also: Seamoth and other Subnautica creatures in the comments

 

Artist: Onion-Oni aka TenTh from Random-tan Studio
Original post: D20 on Tapas (warning: JS-heavy site)

Upscaled by waifu2x (model: upconv_7_anime_style_art_rgb). Original
Unlike photos, upscaling digital art with a well-trained algorithm will likely have little to no undesirable effect. Why? Well, the drawing originated as a series of brush strokes, fill areas, gradients etc. which could be represented in a vector format but are instead rendered on a pixel canvas. As long as no feature is smaller than 2 pixels, the Nyquist-Shannon sampling theorem effectively says that the original vector image can therefore be reconstructed losslessly. (This is not a fully accurate explanation, in practice algorithms need more pixels to make a good guess, especially if compression artifacts are present.) Suppose I gave you a low-res image of the flag of South Korea 🇰🇷 and asked you to manually upscale it for printing. Knowing that the flag has no small features so there is no need to guess for detail (this assumption does not hold for photos), you could redraw it with vector shapes that use the same colors and recreate every stroke and arc in the image, and then render them at an arbitrarily high resolution. AI upscalers trained on drawings somewhat imitate this process - not adding detail, just trying to represent the original with more pixels so that it loooks sharp on an HD screen. However, the original images are so low-res that artifacts are basically inevitable, which is why a link to the original is provided.

24
submitted 3 months ago* (last edited 3 months ago) by [email protected] to c/[email protected]
 

Artist: Onion-Oni aka TenTh from Random-tan Studio
Original post: Knifehead Kaiju on Tapas (warning: JS-heavy site)

Upscaled by waifu2x (model: upconv_7_anime_style_art_rgb). Original

 

Artist: Onion-Oni aka TenTh from Random-tan Studio
Original post: Robot (vacuum) cleaner on Tapas (warning: JS-heavy site)

Upscaled by waifu2x (model: upconv_7_anime_style_art_rgb). Original
Unlike photos, upscaling digital art with a well-trained algorithm will likely have little to no undesirable effect. Why? Well, the drawing originated as a series of brush strokes, fill areas, gradients etc. which could be represented in a vector format but are instead rendered on a pixel canvas. As long as no feature is smaller than 2 pixels, the Nyquist-Shannon sampling theorem effectively says that the original vector image can therefore be reconstructed losslessly. (This is not a fully accurate explanation, in practice algorithms need more pixels to make a good guess, especially if compression artifacts are present.) Suppose I gave you a low-res image of the flag of South Korea 🇰🇷 and asked you to manually upscale it for printing. Knowing that the flag has no small features so there is no need to guess for detail (this assumption does not hold for photos), you could redraw it with vector shapes that use the same colors and recreate every stroke and arc in the image, and then render them at an arbitrarily high resolution. AI upscalers trained on drawings somewhat imitate this process - not adding detail, just trying to represent the original with more pixels so that it loooks sharp on an HD screen. However, the original images are so low-res that artifacts are basically inevitable, which is why a link to the original is provided.

16
submitted 3 months ago* (last edited 3 months ago) by [email protected] to c/[email protected]
 

Artist: Onion-Oni aka TenTh from Random-tan Studio
Original post: The Satellite-girl on Tapas (warning: JS-heavy site)

Upscaled by waifu2x (model: upconv_7_anime_style_art_rgb). Original

This is the Horizon satellite from Random-tan Studio's cybermoe comic Sammy, page 18, prior to remastering.

[–] [email protected] 3 points 3 months ago

Thanks, updated. The picture seems to be Adam Śmietański's concept art based on what is seen of the engine in the official trailer.

[–] [email protected] 6 points 3 months ago

There should be four in the front and four in the back to match the number of hemispheres.

[–] [email protected] 3 points 3 months ago

The posts seem to be getting better lately

view more: next ›