Kling Start & End Frames: First person POV

I thought these images would be great to use this perspective to simulate someone walking through an abandoned area.

Images: Midjourney
Sound: Kling AI Sound
Prompts below: Prompt 1:
First-person POV of walking forward along an abandoned concrete path surrounded by overgrown vines and trees, moving toward the open doorway ahead. Use a whip-pan push-through transition as the camera passes through the doorway, seamlessly revealing the dark, overgrown corridor inside. Maintain the immersive FPV perspective, slight handheld motion, damp atmospheric lighting, and lush green foliage overtaking the walls

Nano Banana Image to Video 3-Step Prompt Worklfow 🍌

Step 1 - Grid theme prompt:

8-panel grid layout:
Panel 1) flowing black velvet cloak on white background,
Panel 2) silver gothic pendant centered on white background,
Panel 3) dark leather boots centered on white background,
Panel 4) candlelit gothic castle hall with stained glass windows,
Panel 5) goblet of blood-red wine on white background,
Panel 6) black raven centered on white background,
Panel 7) ornate dagger centered on white background,
Panel 8) mysterious pale figure portrait on white background. Each item distinct, styled as product photography. Step 2 - Grid to scene prompt:

Remove the grid lines and compose these elements into a dark romantic scene. Show the pale mysterious woman in the hooded black cloak over elegant attire, wearing the gothic pendant necklace and black leather boots. Position her seated in an ornate chair in the candlelit gothic church interior, with the raven perched on the chair's back. She's holding a goblet of red wine in one hand and an ornate dagger rests across her lap. Stained glass windows cast colored light patterns across the scene, creating an atmosphere of elegant menace and aristocratic mystery.

Nano Banana on @lovart_ai is so simple and practical. Select a region, hit tab, and type what you want.

Let the agent do the rest...

Free to use all weekend 🧵 @lovart_ai Workflow:

Select region → Tab → comment → Run.

Results in minutes:

Video Model Comparison: Image to video

I updated an older comparison with a new image and prompt to test in several models:

• Pika 2.2
• PixVerse 4
• Runway Gen-4
• Google Veo 3
• OpenAI Sora
• Kling 2.1 Pro
• Luma Ray2
• Hailuo I2V

I used the same prompt and Midjourney image in each model 2-3 times and chose my favorite results below: Prompt:

Cinematic action shot in the style of an action movie with a drone pilot racing through a forest woodland at noon, navigating between trees. Sunlight streaking through leaves, close front follow angle, dynamic movement, high contrast, intense atmosphere, detailed composition

Original image:

I love using Runway Gen-4 References to continue the story beyond the first frame.

I'll show you how to do it step-by-step below: 1/ I started by placing a character into a Mediterranean scene:

https://x.com/HBCoop_/status/1918352478103294027

The era of lipdub AI is over.

Captions just launched Mirage.

The most advanced AI video model we’ve seen.

Here’s why: Lipdub AI is obsolete:

Most current models only tweak mouth movements while keeping the base video intact.

The result is robotic, awkward, and easy to spot. Mirage doesn’t just move lips. It generates actual reality.

How to train a FLUX LoRA on an illustrated character sheet to generate realistic, consistent characters.

Workflow below: I started with this character sheet I generated with FLUX on @krea_ai.

I used @CleanShot to take screenshots of each pose, and upscaled them:

🔳 Quick prompt tips: Adding 'telephoto shot' with 'soft focus' creates natural bokeh effects.

Notice how light particles blur naturally in the background while maintaining sharp detail on the subject?

Prompt below:

@Hailuo_AI @Hailuo_AI "Extreme telephoto shot slowly transitioning from soft focus to sharp detail on the panther's eye. Environmental particles create magical atmosphere as golden hour lighting wraps the creature's face. Dreamlike depth of field transitions, natural wildlife movements"

@Kling_ai

FLUX LoRA Resources | Part 4

Mega thread👇 1. PuLID for FLUX: Tuning-free ID customization with FLUX-dev

https://x.com/Gradio/status/1836051339178082485

🔎Midjourney vs. FLUX

I was surprised at the similarity between these images generated from the same prompt.

But there are noticeable differences in style, color, depth of field, and mood.

I'll show you how I got to this comparison👇 1. I used Midjourney sref 1542881645 with this prompt, for the original image:

"photorealistic, neon body tattoos, in the style of cyberpunk dystopia, a man, cityscape"

It's an old prompt I've been using for about 1 year.

This is the Midjourney result:

How to train a FLUX LoRA on your own images using @FAL

Full tutorial and video 👇 @FAL 1. Collect 10-20 selfies or photos of yourself in different poses and backgrounds.

2. • Use the same prefix (e.g., "Heather1, Heather2") for easy retrieval.

3. Optional: Put your images in a folder or zip file.

FLUX LoRA Resources | Part 3

Mega Thread👇 1. Run multiple FLUX LoRAs to combine styles or use on more than one subject in an image:

https://x.com/venturetwins/status/1825924552817520896

I just fine-tuned a Flux.1 Dev LoRA with my own images on @Astria_AI.

The results are incredible!

Workflow and tutorial👇 @Astria_AI First, I'll answer the question I posted yesterday.

Image 3 is the real me 😊

Most of you answered correctly!

https://x.com/HBCoop_/status/1826039100060635285

FLUX.1 LoRA Resources | Part 2

Mega Thread 👇 1. Train on your own image - Replicate workflow by @mreflow

https://x.com/mreflow/status/1823778726330745256

FLUX LoRA Resources

Mega Thread👇 1. FLUX.1 LoRA Training @mreflow

https://x.com/mreflow/status/1823234197114499241

The magic of this FLUX prompt?

It’s all in the details!

By specifying a futuristic AI lab, holographic displays, and advanced tech, we got a highly realistic, immersive scene.

The focus on interactions and sleek design made it perfect for animation with stunning VFX 👇 I used @FAL's FLUX Realism LoRA model to generate the image below.

Prompt in ALT:

✨FLUX: Cosmetic Ad

8-Step workflow and tutorial below: 1. I used the Flux Prompt Enhancer Glif by @angrypenguinPNG to generate a detailed prompt for a "UI design cosmetic brand homepage"

2. I dropped the prompt into @FAL's FLUX Realism LoRA.

3. I asked Claude for 3 video prompts to animate the images with Gen-3:

✨Another new AI video tool: Veo by @GoogleDeepMind

60-second clips at 1080p like this:

(Link & more examples below) @GoogleDeepMind Google DeepMind released unedited, raw videos to show the capability of Veo:

1.

https://twitter.com/GoogleDeepMind/status/1790435828709126264

✨TUTORIAL✨

Runway's Generative Audio tool gives you more options:

• Text-to-voice or voice cloning
• Animate & lip sync static images
• Add audio & lip sync effects to videos

Upload audio files, generate with a preset voice, or record audio through the mic:

@runwayml @runwayml How to make a Lip Sync video:

Go to Generative Audio > Text to Speech tool on Runway

Get your audio ready by uploading an existing file, recording a new clip, or generating text-to-speech audio

Upload or select the image/video of the face you want to animate

Click Generate:

🔹PAPER QUILLING KIRIGAMI🔹

I tested some of my old Paper Art prompts in Midjourney V6.

Here's some of my amazing results with prompts so you can try them too: 🔹Prompt: Paper quilling kirigami, 3D, stunning close-up portrait of a woman with flowers and colors swirling around her, taken from the neck up, lovely

🟨 REFLECTIONS: MIXED MEDIA COLLAGE 🟨

"Mixed media collage" means that the image should be made up of different materials, such as paper, fabric, or photographs, that are arranged together to create a cohesive whole.

I'll show you 3 examples of AI-generated collages: 👉Prompt: Mixed media collage showing an airplane flying over a reflective lake with sunlight and shadows from afternoon sunny day