Moon 🎑
@eveilore
Followers
418
Following
11K
Media
425
Statuses
2K
AI x Art = 🪻Infinite Possibilities 🌠 | Robotics, Futurism, and Exploring Beyond the Horizon, occasionally posts something important on latest AI 💭
Joined June 2020
A new image model, Seedance 4.0, has been released by ByteDance, featuring strong image editing capabilities like Nano Banana. It can generate image or can also edit image using text instructions in image annotations or by mixing groups of images.
0
1
4
I try to create the same with Seedance 4. Turns out, It also follows the annotation instructions in the image. These results are very good. 👏
This is the image I’m using on @Adobe Firefly Board with Nano Banana (Gemini 2.5 Flash). My suggestion: always use annotations on each object, because Nano Banana works way better when every object is clearly labeled for the character we’re developing.I’ll drop the prompt and
1
0
1
Something's up from Baidu on tomorrow. Ernie AI model upgrade, PaddlePaddle, Comate code assistant and more! ❤️🔥
Wave Summit 2025 kicks off tomorrow with exciting updates on ERNIE, PaddlePaddle, Baidu Comate, and more. But that’s not all—for the first time ever, the team behind ERNIE will go live right after to share more on our upgraded model. Tune in on X or YouTube to be the first to see
0
1
2
ByteDance officially launched dreamina Image version 4.0. debuting multimodal image generation. The model integrates three core capabilities: text-based image generation, image editing, and group image generation, and supports flexible adjustment of image details through natural
0
0
5
. @NanoBanana create an image of childlike inkjet art illustration uses light beige as the canvas and uses highly saturated colors (bright red, bright yellow, lake blue, grass green, and pink purple) to paint irregular, translucent fruits and vegetables. The edges of the fruits
0
0
1
RAI Institute is building robots called Ultra Mobile Vehicle (UMV) that think, plan, and move like athletes robots that combine the mobility of parkour with the intelligence to perceive, plan, and navigate any terrain, no matter how complicated.
2
0
3
Meta last model release was 4 months ago in April. No matter how much they talk, they can't help Meta come up with Llama 5 or even 6. Because basically all the big AI companies have their own specific AI products or models, and have bet on it as unicorn. Open AI has the GPT
0
0
1
Qwen Image generation is just fine. You don't need to pay for Midjourney to generate random art, there are lots of free image generators with good image output.
0
0
1
Flying vehicles into the future 🛸 using Baidu Steam Engine Video model.
0
0
2
Microsoft's open source model tops the trending list of Huggingface. VibeVoice: A cutting-edge open source text-to-speech model. The model can synthesize up to 90 minutes of speech with up to 4 different speakers.
0
1
10
Be Hyper-Specific with much details as possible, Provide Context and Intent by explaining the purpose of the image, Iterate and Refine, Use Step-by-Step Instructions breaking complex scene in steps, Use Semantic Negative Prompts, Control the Camera.
1
0
1
Prompts for editing images : Adding and removing elements, Inpainting (Semantic masking) to edit a specific part of an image only, Style transfer, Advanced composition: Combining multiple images, and High-fidelity detail preservation.
1
0
0
Describe the scene, don't just list keywords. The model's core strength is its deep language understanding. A narrative, descriptive paragraph will almost always produce a better, more coherent image than a list of disconnected words.
1
0
0