What AIs are you using lately to create videos with a fixed character?

Advertise on BHW
Depending on what you mean, im not super into it, but i hear persyn.ai is good, or just Gemini, their new version of nano banana is crazy for image generation, aswell as their VEO 3.1 model for videos, and there is a chinese one which i completely forgot the name of that requires a chinese number to sign up to. but for generous free tiers id go Nano Banana for the image generation and VEO 3.1 for the video generation.
 
Honestly, Midjourney with the --cref parameter is the best thing right now for keeping a character consistent. It is very faithful to the original face and does not mess up the features as long as your prompt is clear.

If you want professional control then look into Stable Diffusion with IP-Adapter or FaceID. Everything else just hallucinates too much and changes the character every time you change the scene. Stick to MJ if you want it fast and clean.
 
For actual video MJ --cref gets you consistent stills but you still need a video step. Kling handles the image-to-video part and keeps the character reasonably stable across cuts if you feed it the same reference. HeyGen is worth looking at if the use case is a talking character — it's built specifically for that and doesn't drift the way general video models do.
 
Honestly, after working on many projects, i find Kling AI (version 3.0) and Runway Gen-3 to be the two apps that maintain character stability the most. Kling feels very stubborn, with characters showing minimal distortion even during strong movements, especially when creating videos up to 3 minutes long. If you want the highest fidelity to the original image, use Luma Dream Machine's Reference-to-Video feature. It adheres to the original facial structure and clothing incredibly well
 
Lately I see a lot of people using Runway and Kling for that. Consistency is improving, but most tools still need some tweaking to keep the character accurate. Usually good prompts and reference images make the biggest difference.
 
I've tried several tools, but the most stable for maintaining character consistency is Runway. If you use the right reference (original image, clear prompt) the character is less "distorted" than with other tools. However, no AI tool can perfectly preserve the original image 100%, manual adjustments are still needed to ensure accuracy and naturalness. You can try it out to see if it suits your job requirements.
 
Character consistency is two separate problems — face/body and motion. For face/body across frames: Midjourney with --cref or character reference, or Flux LoRA trained on 15-20 reference shots gives the most stable look. For motion: Runway Gen-3 with the locked-character workflow (img2vid keeping the seed frame) holds up better than VEO for clips under 6 seconds. Combine both — generate consistent stills first, then animate each as separate clips and stitch.
 
for fixed characters, workflows using reference images and character locking usually work best. lots of people are using tools like Runway, Pika, Kling AI, and Luma AI for consistent AI characters lately. Kling and Runway usually is the most for keeping character appearance stable when using strong reference images and consistent prompts
 
Advertise on BHW
You must log in or register to reply here.
Sign up now!

Main Menu

Home Main Forum List Black Hat SEO White Hat SEO BHW Newbie Guide Blogging Black Hat Tools Social Media Downloads

Marketplace

Account Selling Content / Copywriting EBooks / Methods Hosting Misc Proxies For Sale SEO - Link building SEO - Packages Social Media Social Media - Panels Web Design More

Making Money

Affiliate Programs Hire a Freelancer IM Journeys Making Money Pay Per Click Site Flipping Want to Buy

BlackHat World

Dispute Resolution Forum Suggestions & Feedback Introductions Lounge
Back
Top