And has this guy never seen a chorse before?
(Okay, I got nothing after that one. Based on my limited knowledge of French, this is appears to be a terrible translation issue- tree = arbre, horse = cheval, house = maison.)
G für Gadze
That book is a translation nightmare…
Tree = árvore Horse = cavalo Vaca = cow
There are a lot more. Those are English to Portuguese translations and there are more. I’d risk the original language of the book is heavily Latin based.
Not sure if it’s originally Portuguese, Spanish, or French but yeah, definitely seems like it was originally some romantic language originally and they just translated it one to one to English without considering the context.
For example in Spanish, E for Estrella (star) makes sense.
Can confirm these three also work in french and Spanish!
Every single one works in French.
It’s an image translated by an AI and used for internet points. Zoom in to the text to see why.
Yeah, this is ai slop
HELLCOPTER
Gateau = cake
it’s probably translated from another language yeah, could be French or Spanish
- Tree in French: Arbre; in Spanish: árbol
- Horse in French: Cheval; in Spanish: caballo
Is it just me or are the images and text AI generated?
It’s Google Translate, from French I think.
Tree= Arb Horse = Cheval Cow = vache
I can’t see the while thing while I comment, but I know a Translate image when I see it.
It seems people are blind to AI now, seeing people having discussion over this AI slop makes it time to add “braindead society theory” to the already existing “dead internet theory”
100%, the text was an instant giveaway and the disagreement makes me feel like I’m living in a Truman Show simulation sometimes. Also, the monkey is actually a nightmare creature and the tiger is straight from the Walking Dead.
People are discussing the word “horse” as if it’s present in the image, but it’s just some garbled shapes, not even letters, except for “h” and two "o"s. WTF?
I think it’s AI upscaling
that’s likely it yes
Yeah. What kind of GenAI would be so shitty to render something with so many artifacts, yet coherent enough to render 24 words that perfectly map to their direct French translation? But somehow the pictures are half jumbled to the point that the picture of a tail looks like a circle? Which is the opposite way GenAI normally jumbles things, text is always the first to become undecipherable.
The only way for this to be GenAI would be with close supervision, it’s not impossible but at that point it would have been much less effort for a much better result to edit English text onto an actual French children’s book.
Anyway who gives a shit but the superior attitude of the people here who think they are so clever pisses me off lol
very clearly yes
Ah yes, P for Phish, classic kindergarten stuff
Yote is the past tense of yeet
How are people not noticing this “book” is entirely AI generated with all the fucked up non-letters in the words?
But Abazz said his mom bought that book. He wouldnt lie to me :(
Because it isn’t. They took the picture with the book on their lap, after covering the words with white out and writing the ones we see over it.
Except there’s no indication of whiteout (it’s usually pretty obvious) and all the fucked up letters look exactly the same (such as e looking like an O with a dot inside.
It’s much more likely they just put the image into some AI thing to translate the words.
G: Cake M: House V: Cow
I didn’t check all of them.
This is AI though, or am I being whooshed
I don’t think it’s AI. I think it’s just an upscaling method that REALLY doesn’t like text. I’ve seen it around on Lemmy often
it fucks it up so much I’m pretty sure it’s ai upscaling
not sure if the book is AI or if there’s AI uprezzing, check the letters in the microblog too.
Vaca is cow so that explains v
Vache is cow in French.
TIL
Fetchez la vache!
Either the whole image is AI gen or someone used AI gen to make a childrens book.
I think its the former and the whole post is just bullshit.
I also believe it to be the former, however I love the idea of the latter as it would be a great example of how not to use image generation models.
If it’s AI generated, it’s a hell of a coincidence that the letters all match the French words for those animals.
All the words are 100% generated unless someone went to the trouble of making a font that looks exactly like AI generated text. The pants under the book are also just nonsense blobs.
Again, that is one hell of a coincidence.
I actually think you are both right… But it’s likely some one is using an AI translator that is trying to reconstruct the original picture with the translated words substituted in the original words place.
no, it’s gotta be all ai. look at the image for helicopter.
Yeah it probably recreated the whole image
unless someone went to the trouble of making a font that looks exactly like AI generated text.
Can I just say that’s a hilariously evil idea to fuck with people and if I knew more about font formats I’d be tempted to give it a go.
It’s AI. Look at the word that’s supposed to be “Horse”.
It’s probably in-place translation using AI for a French book
I think we have the solution here.
I guess that app also fucks with everything else in the image tho, not just text.
Could very well be ML repainting, “draw this image with English text”
The book is 1000% AI generated. The text and images are blurry mid journey nonsense. I mean, “helicopter” is a fucked up looking plane ffs.
However, it would be a massive coincidence that all the words have a French translation that fits the letter. The OP could’ve done that intentionally, but that seems like a lot more effort for an AI generated post
Given the perfect grid pattern and a certain kind of coherence this kind of ML doesn’t usually preserve it’s much more likely somebody cut and paste the individual images into an ML based image generator to repaint them with English text
Stupid? Yes. They could have just taken the text alone into an LLM, or better yet regular translation program. But since when was the kind of people who blindly rely on ML smart?
It looks to me like google translate text over an image of a French book. The text is garbled, but it’s close enough and the words are accurately translated.
The whole screenshot is AI upscaled. It’s obvious if you zoom into the first emoji.
“Helloopter”
It’s just the shitty resolution, but still.
Can we acknowledge how weird it is that the word they chose for U was “Universe”. Seems a bit grandiose for a letter book for young children.
“A is for apple. B is for balloon. C is for THE COSMIC VOID OF INFINITE SPACE”
D is for dice, E is for elephant, F is for the fascists threatening our democracy
“H is for: Horrors beyond human comprehension”
M is for mesothelioma
