GCVision

[GCVision @randomgreekuser]

@gourpan Τι εννοείς θα έχουν αποκλείσει ήδη ένα φαβορί;
Με τον ηττημένο αυτού του ζευγαριού παίζουμε, όχι με τον νικητή.

[GCVision @randomgreekuser]

@Sophiaapmo This AI. Many details of the dog are changing between each shot

[GCVision @randomgreekuser]

@Den13_1908 Ναι, πέρα από την κατάκτηση του Europa, φτιάχνεις και την επόμενη χρονιά σου μιας και εξασφαλίζεις απευθείας συμμετοχή στο επόμενο Champions League χωρίς να χρειάζεται να τρέχεις το καλοκαίρι για προκριματικά, θα χεις και δελεαρ για να φέρεις μεγαλύτερους παίκτες με μεταγραφή

[GCVision @randomgreekuser]

@nikkitheninjax @DavidAriew As I mentioned earlier, it’s not good yet, probably because there aren’t many audio visualizers in the training data. The audio reactive effect works better when you prompt it to generate people dancing on a stage with strobing lights.

[GCVision @randomgreekuser]

@ostrisai Will it be possible to train IC-LoRA?

[GCVision @randomgreekuser]

@ostrisai @Lightricks Do you plan to add support for LTX-2 LoRA training to AI-Toolkit?

[GCVision @randomgreekuser]

@LTXStudio Is it possible to provide your own audio and have LTX handle the lip-sync?

Regarding the open-source release, which variants of the model will be made available?

[GCVision @randomgreekuser]

@naipoioseinai1 Μην παίρνεις στα σοβαρά τα διάφορα εργαλεία που υπολογίζουν κατά πόσο είναι AI-generated κάποιο περιεχόμενο

[GCVision @randomgreekuser]

@paoworldwide13 Μια γρήγορη σύγκριση μεταξύ Dessers, Ιωαννίδη, El Kaabi και El Arabi σε γκολ και ασίστ ανά αγωνιστικά λεπτά #paofc

[GCVision @randomgreekuser]

@deksixeripae Ετοιμάζεται να την πυροδοτήσει

[GCVision @randomgreekuser]

@KampanakiD Έχει κλείσει σίγουρα εκεί;
Προς το παρών δεν βλέπω να χει κάνει κανένα like σε post της Girona

[GCVision @randomgreekuser]

@es_pikopikola In a mirror, you can see yourself and adjust your expression. Similarly, when you wear headphones and try to sing without hearing your own voice, you’re likely to sing out of tune

[GCVision @randomgreekuser]

@mgabriel93 @StevenZhang66 @Alibaba_Wan You'll also need to prepare a mask video. At the keyframe positions (1, 15, 20, etc.), the frames should be solid black; all other frames should be solid white.

Currently, only the 1.3B model is available. The 14B model, once available, should offer better quality.

[GCVision @randomgreekuser]

@mgabriel93 @StevenZhang66 @Alibaba_Wan You can do it with VACE. First, prepare a control video by placing the image keyframes at the corresponding frames of the video — for example, frames 1, 15, 20, etc. All other frames should be filled with solid gray (RGB 127)

[GCVision @randomgreekuser]

@StevenZhang66 @Alibaba_Wan I'd love to see a higher native framerate in future versions. I know you can use frame interpolation to increase it, but in fast-moving scenes like explosions or fireworks, it often causes warping or visual distortions, so it doesn't always work well.

[GCVision @randomgreekuser]

@RedmondAI @cocktailpeanut It modifies only the inpainted region based on your prompt and the references images (if you use any).

You you can inpaint the first frame with an image model like Flux, and let VACE generate the rest frames of the inpaint regions based on the first frame.

[GCVision @randomgreekuser]

@cocktailpeanut According to a GitHub comment, an official model with Start/End frame support will be released. Perhaps it will work better with this model.
https://t.co/dR67Bc9lmN

[GCVision @randomgreekuser]

@cocktailpeanut Here is my result with the following prompt: A timelapse of pale, uncooked croissant dough gradually rising and transforming into perfectly baked, golden-brown croissants. Steam escapes as the dough expands and changes color, creating a smooth and satisfying transition

[GCVision @randomgreekuser]

@cocktailpeanut @deepbeepmeep I installed it with Pinokio but the "Gen Textured Shape" button" is missing