Why AI Video Still Gets Hands and Faces Wrong (And How to Fix It)
Hands and faces are the two things AI video breaks most often, and the two things a viewer notices fastest. It happens because the model has no memory between frames and has to resolve enormous structural detail, joints, fingers, facial geometry, packed into a small part of the image every time it generates. This guide covers why that happens and what fixes it. For distortion more broadly, see our companion guide on avoiding distortions in AI video.
Why Hands and Faces Break More Than Anything Else in the Frame
Hands pack an enormous amount of detail into a small area. Joints bend in specific directions, fingers overlap and curl in ways that are structurally complex even for a human artist to draw. The model has to get the finger count, the joint angles, and the overlap right all at once, in a fraction of the frame. In Cinema Studio, setting the lens, focal length, and aperture based on how much of the frame the hand occupies fixes that density problem directly.
Training data has real gaps around hands specifically. Hands appear gripping, pointing, resting, gesturing, often partially occluded by an object or motion blur, so the model has seen fewer clean examples of a given hand position than it has of a face looking forward at a camera. Uploading a reference image of the exact hand position alongside the character reference on Seedance 2.0 closes that gap directly.
An overloaded prompt gives the model too many things to resolve at once. Describing the face, the hand gesture, the object, the lighting, and the camera all in one dense paragraph splits the model's attention, and hands and faces are the first details to degrade under that load. Moving the camera and lighting into Cinema Studio's settings takes that weight off the prompt, so it can stay focused on the hand or face alone.
5 Tips to Avoid Broken Hands and Faces
Fix
What you do
What it removes
Replace face description with a trained identity
The model's need to reinterpret "a woman with dark hair" fresh each time
Use Cinema Studio to control framing
Set lens, focal length, and aperture based on how much of the frame the hand or face occupies
Ambiguity about how much structural detail the model needs to resolve at that scale
Describe start and end states, not motion
State the hand's exact position before and after, not the gesture itself
The open-ended motion path the model would otherwise have to invent
Choose a model built for multi-reference input
Upload a reference image of the exact hand position alongside the character reference on Seedance 2.0
The need to approximate a gesture from text alone
Keep the prompt focused on the hardest detail
Describe the hand or face first, with the most specific language, before secondary details
Competition for the model's attention from background, lighting, and wardrobe details
Step 2: Set the Cinema Studio parameters for the shot. Choose the genre, lighting preset, lens, focal length, and aperture based on how much of the frame the hand or face occupies. A close macro shot on a hand needs a different depth of field than a wide shot where the face is a smaller part of the frame.
Step 4: Write the prompt around start and end states. Describe the hand or face position at the beginning of the shot and the position at the end, rather than describing the motion itself. Keep this description first in the prompt, ahead of secondary details like background and wardrobe.
Step 5: Generate and check the hardest detail first. Review the hand or face before checking anything else in the frame. If it holds up, the rest of the shot is very likely fine. If it does not, adjust the reference image or the start/end description before touching anything else.
Settings Reference
Tool
Key capability
What it fixes for hands and faces
Trains identity from 20+ reference photos
Face reading as a different person shot to shot
Works across Kling 3.0, Veo 3.1, Seedance 2.0, WAN 2.6
Same face holding across every model, not just one
Persists across Cinema Studio, Marketing Studio, LipSync Studio
No re-uploading a face reference per tool
Cinema Studio
Per-shot camera control (not text-described)
Framing decisions that control how much hand or face detail the model must resolve
6 lens options, 5 focal lengths, 3 aperture settings
Depth of field and scale, so a close macro hand shot gets a different setup than a wide face shot
10 camera movement styles
Predictable motion paths, reducing the ambiguous mid-gesture guessing that produces extra fingers
Seedance 2.0
Up to 12 reference inputs per generation
Real visual material for a hand position or face, instead of a text approximation
What Happens Without a Trained Identity
The Same Face Across Every Scene
With a trained AI character in place, everything else around them becomes flexible. Switch models, change the location, put them in a completely different action or scene, and none of it touches the one thing that actually needs to stay fixed: the face and the body. Below is that in practice. Same trained character, several different prompts, all run through Seedance 2.0 and Cinema Studio. Here's what came out across each one.
The Fix Is Always the Same Shape
Hands and faces break for structurally different reasons. Hands fail because of raw detail density packed into a small part of the frame. Faces fail because the model has no memory of what it generated a moment ago. But every fix in this guide solves both problems the same way: give the model something concrete to hold onto instead of leaving it to infer.
A trained identity replaces a guessable text description with a fixed object the model applies rather than reinvents. A reference image replaces an ambiguous gesture with a real visual target instead of a guess. Start and end states replace an open-ended motion with two fixed points to interpolate between, rather than an unbounded path the model has to invent on its own. Framing choices in Cinema Studio decide upfront how much structural detail the model actually has to resolve, instead of leaving that decision to chance. None of these fixes make the model smarter. They all just remove the specific ambiguity that was causing it to guess wrong in the first place.
Why AI Video Still Gets Hands and Faces Wrong (And How to Fix It)
Got any questions left?
by TAVOA AI