Загрузка видео...

Не удалось загрузить видео

На главную

From LA to New York, the model continuously impresses me. This time utilizing an Image To World, the consistency in aesthetics from the input image is remarkable.

TheoMediaAI's profile picture

Theoretically Media

12,428 subscribers

34,467 просмотров • 6 месяцев назад •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

Комментарии: 0

Нет доступных комментариев

Здесь появятся комментарии из оригинального поста

Похожие видео

Gemini 3.1 Pro is the best model in the world for going from image to code. This task is basically solved now, kind of crazy. The model is now available in MagicPath.
0:27

Gemini 3.1 Pro is the best model in the world for going from image to code. This task is basically solved now, kind of crazy. The model is now available in MagicPath.

Pietro Schirano

219,384 просмотров • 5 месяцев назад

Nano Banana is fantastic for consistency. Created this entire coffee shop advert just from a single image of a logo. Here are the prompts. Original image from Ideogram. Nano Banana to reimagine the logo in new places. Runway Gen-4 Turbo to animate to video. Incredible.
0:26

Nano Banana is fantastic for consistency. Created this entire coffee shop advert just from a single image of a logo. Here are the prompts. Original image from Ideogram. Nano Banana to reimagine the logo in new places. Runway Gen-4 Turbo to animate to video. Incredible.

Jerrod Lew

19,843 просмотров • 11 месяцев назад

My favorite demo of the new gpt-realtime model from Matthieu LC -- Shoggoth Mini using Realtime API with image input
2:45

My favorite demo of the new gpt-realtime model from Matthieu LC -- Shoggoth Mini using Realtime API with image input

Peter Bakkum

67,871 просмотров • 11 месяцев назад

TurboEdit Instant text-based image editing discuss: We address the challenges of precise image inversion and disentangled image editing in the context of few-step diffusion models. We introduce an encoder based iterative inversion technique. The inversion network is conditioned on the input image and the reconstructed image from the previous step, allowing for correction of the next reconstruction towards the input image. We demonstrate that disentangled controls can be easily achieved in the few-step diffusion model by conditioning on an (automatically generated) detailed text prompt. To manipulate the inverted image, we freeze the noise maps and modify one attribute in the text prompt (either manually or via instruction based editing driven by an LLM), resulting in the generation of a new image similar to the input image with only one attribute changed. It can further control the editing strength and accept instructive text prompt. Our approach facilitates realistic text-guided image edits in real-time, requiring only 8 number of functional evaluations (NFEs) in inversion (one-time cost) and 4 NFEs per edit. Our method is not only fast, but also significantly outperforms state-of-the-art multi-step diffusion editing techniques.
0:22

TurboEdit Instant text-based image editing discuss: We address the challenges of precise image inversion and disentangled image editing in the context of few-step diffusion models. We introduce an encoder based iterative inversion technique. The inversion network is conditioned on the input image and the reconstructed image from the previous step, allowing for correction of the next reconstruction towards the input image. We demonstrate that disentangled controls can be easily achieved in the few-step diffusion model by conditioning on an (automatically generated) detailed text prompt. To manipulate the inverted image, we freeze the noise maps and modify one attribute in the text prompt (either manually or via instruction based editing driven by an LLM), resulting in the generation of a new image similar to the input image with only one attribute changed. It can further control the editing strength and accept instructive text prompt. Our approach facilitates realistic text-guided image edits in real-time, requiring only 8 number of functional evaluations (NFEs) in inversion (one-time cost) and 4 NFEs per edit. Our method is not only fast, but also significantly outperforms state-of-the-art multi-step diffusion editing techniques.

AK

16,062 просмотров • 1 год назад

Imagine a world where every image has a ZK verified checkmark, proving that the image came from an authentic camera, not generated by an AI model. threadguy wants to know, "Who's building this? I want to invest."
1:38

Imagine a world where every image has a ZK verified checkmark, proving that the image came from an authentic camera, not generated by an AI model. threadguy wants to know, "Who's building this? I want to invest."

Succinct

19,382 просмотров • 11 месяцев назад

Generative image to scene is getting incredibly good. This is pure 3D without v2v generated from the image in the reply ...
0:23

Generative image to scene is getting incredibly good. This is pure 3D without v2v generated from the image in the reply ...

martin_casado

57,089 просмотров • 11 месяцев назад

Excited to share Muse Spark! It's a strong natively multimodal model with many surprising properties that emerged. Here, the model is able to use Python tools to make a playable Sudoku game on the web from an image input of the board. ✨
0:42

Excited to share Muse Spark! It's a strong natively multimodal model with many surprising properties that emerged. Here, the model is able to use Python tools to make a playable Sudoku game on the web from an image input of the board. ✨

Matt Deitke

29,459 просмотров • 3 месяцев назад

Halfmoon is Reve Image — and it’s the best image model in the world 🥇 (🔊)
0:45

Halfmoon is Reve Image — and it’s the best image model in the world 🥇 (🔊)

Reve

546,809 просмотров • 1 год назад

📢ProcGen3D: Learning Neural Procedural Graphs for Image-to-3D Reconstruction Xinyi Zhang learns neural procedural graphs to generate high-fidelity 3D - MCTS-guided sampling maintains consistency with the input image, even from real images! Check it out:
0:38

📢ProcGen3D: Learning Neural Procedural Graphs for Image-to-3D Reconstruction Xinyi Zhang learns neural procedural graphs to generate high-fidelity 3D - MCTS-guided sampling maintains consistency with the input image, even from real images! Check it out:

Angela Dai

14,639 просмотров • 8 месяцев назад

Flux Kontext is now live in Hedra! Now you can use an image-to-image model to edit your images while keeping character consistency. Try it in Hedra Studio.
0:42

Flux Kontext is now live in Hedra! Now you can use an image-to-image model to edit your images while keeping character consistency. Try it in Hedra Studio.

Hedra

94,348 просмотров • 1 год назад

We vibe coded this demo in Google AI Studio to show Nano Banana 2's real-world understanding. With each frame in this environment, the model sees only the previous image, and is prompted to imagine what happens next. The consistency is incredible!
0:36

We vibe coded this demo in Google AI Studio to show Nano Banana 2's real-world understanding. With each frame in this environment, the model sees only the previous image, and is prompted to imagine what happens next. The consistency is incredible!

Demis Hassabis

69,141 просмотров • 5 месяцев назад

Today, we’re launching Reve 2.0, the best 4K image model in the world. We invented a new way to generate and edit any image using precise layouts. For the first time, it’s possible to create images you can touch.
0:54

Today, we’re launching Reve 2.0, the best 4K image model in the world. We invented a new way to generate and edit any image using precise layouts. For the first time, it’s possible to create images you can touch.

Reve

11,930,596 просмотров • 2 месяцев назад

Seedream 5.0 Pro is now in ComfyUI via Partner Nodes. ByteDance's latest image model brings: → Character & product consistency across edits → Precision region editing → Infographics, flowcharts & structured layouts → In-image text across 14 languages One model. Text-to-image, editing, and multi-image inputs. Click the link below for the workflow 👇
0:32

Seedream 5.0 Pro is now in ComfyUI via Partner Nodes. ByteDance's latest image model brings: → Character & product consistency across edits → Precision region editing → Infographics, flowcharts & structured layouts → In-image text across 14 languages One model. Text-to-image, editing, and multi-image inputs. Click the link below for the workflow 👇

ComfyUI

14,756 просмотров • 26 дней назад

F22 Fighter Pilot simulation with 1 input image with the incredible new #Genie3 from Google GoogleDeepMind
0:59

F22 Fighter Pilot simulation with 1 input image with the incredible new #Genie3 from Google GoogleDeepMind

GenMagnetic

60,948 просмотров • 6 месяцев назад

📢WorldAgents: 3D worlds only from 2D image models - without any training! We propose an agentic approach with a Director (VLM) to plan the scene, a Generator (Flux or NanoBanana) for new views, and a Verifier (VLM) for selection / 3D consistency. -> High-fidelity 3D worlds from a single text prompt. What's remarkable: our agents find consistent views from 2D image models to obtain 3D-consistent worlds; this shows that image models contain world priors - agents just need to find them! Great work by Ziya Erkoç Angela Dai
0:25

📢WorldAgents: 3D worlds only from 2D image models - without any training! We propose an agentic approach with a Director (VLM) to plan the scene, a Generator (Flux or NanoBanana) for new views, and a Verifier (VLM) for selection / 3D consistency. -> High-fidelity 3D worlds from a single text prompt. What's remarkable: our agents find consistent views from 2D image models to obtain 3D-consistent worlds; this shows that image models contain world priors - agents just need to find them! Great work by Ziya Erkoç Angela Dai

Matthias Niessner

19,004 просмотров • 4 месяцев назад

For this video created from an image in Grok Imagine, I combined two techniques: - the image editor from Grok Imagine - starting from a frame of a video to create other videos.
1:00

For this video created from an image in Grok Imagine, I combined two techniques: - the image editor from Grok Imagine - starting from a frame of a video to create other videos.

Déborah

18,012 просмотров • 6 месяцев назад

"AI telephone" starts with an image -> GPT describes the image -> DALL-E makes an image from that description -> GPT describes the image -> DALL-E makes an image from that description, etc.... Experiment from my artist residency at OpenAI
1:04

"AI telephone" starts with an image -> GPT describes the image -> DALL-E makes an image from that description -> GPT describes the image -> DALL-E makes an image from that description, etc.... Experiment from my artist residency at OpenAI

Alexander Reben

39,236 просмотров • 2 лет назад

My first Google Veo 3.1 animated short 🤖 Made from one single image This is Bloom Using Scenebuilder and the New Insert function I was able to extend the scene and prompt new scenes from one hero image.
1:49

My first Google Veo 3.1 animated short 🤖 Made from one single image This is Bloom Using Scenebuilder and the New Insert function I was able to extend the scene and prompt new scenes from one hero image.

Dave Clark

167,731 просмотров • 9 месяцев назад

Image analysis is now live for Pro users When you drag and drop an image into the input field Venice now auto-selects Qwen 2.5 VL 72B to analyze your image
0:21

Image analysis is now live for Pro users When you drag and drop an image into the input field Venice now auto-selects Qwen 2.5 VL 72B to analyze your image

Venice

53,759 просмотров • 1 год назад

Underrated: seed v0 generations from image models. I gave a source image from the Black Forest Labs FLUX.1 model. It implemented animations, responsiveness, copy, gradients. The phone in the image was implemented in pure CSS / Tailwind.
0:28

Underrated: seed v0 generations from image models. I gave a source image from the Black Forest Labs FLUX.1 model. It implemented animations, responsiveness, copy, gradients. The phone in the image was implemented in pure CSS / Tailwind.

Guillermo Rauch

19,555 просмотров • 1 год назад