Prompting
Describe the final image, not the edit command
Prompting Ray 3.2 starts with the input mode. For text-to-video, the prompt should establish the scene, subject, motion, camera behavior, and visual style. For image-to-video, the image anchors the subject or composition, while the prompt describes the motion and finished video look.
Your prompt should describe the target end state. Do not write loose edit commands. Describe what the video should look like, how it should move, and which visual qualities should remain consistent across the generated clip.
Instead of
Change the sky to purple.
Use
A purple sky over the mountain landscape.
Instead of
Remove the people from the street.
Use
An empty cobblestone street at dawn.
Instead of
The colors shift throughout the shot.
Use
Saturated teal-and-orange color palette.
Ray 3.2 also responds better to positive phrasing. Avoid no, not, without, devoid of, or missing. For removal, describe the empty result. For style direction, describe the finished style. For material or wardrobe changes, describe the final look.
Example
Preserve all other elements including subject identity, pose, and lighting.
Enhance
Use Enhance as a workflow choice
Enhance is on by default. When enabled, Ray 3.2 can rewrite a rough instruction with model-friendly generation guidance before the run. This helps turn a loose idea into a clearer prompt for T2V or I2V creation.
Use Enhance when you have a simple idea, rough instruction, or prompt that has not been optimized yet. Turn Enhance off when the prompt has already been written carefully and you want the model to receive the exact wording.
Simple rule
- Rough prompt or fast exploration: Enhance on.
- Carefully written production prompt: Enhance off.
Templates
Prompt structures to adapt
Use these as sentence shapes, not scripts. The important pattern is target description first, preservation cue second.
Material swap
Structure
A [subject] made of [target material], with [surface qualities]. Preserve all other elements including subject identity and pose.
Example
A sports car made of brushed black titanium, with subtle satin reflections and crisp panel lines. Preserve all other elements including subject identity and pose.
Environment swap
Structure
[Subject] in [new environment], [time of day], [weather]. Preserve subject identity, wardrobe, and pose.
Example
A runner on a quiet Tokyo street at night, wet pavement, soft neon reflections, light rain. Preserve subject identity, wardrobe, and pose.
Style transfer
Structure
[Scene description] rendered as [target style]. Preserve composition and motion.
Example
A city street scene rendered as hand-painted anime background art, clean linework, soft cel shading, graphic color blocks. Preserve composition and motion.
Wardrobe or product change
Structure
[Subject] wearing [new wardrobe / holding new product]. Preserve identity, environment, pose, and lighting.
Example
A presenter wearing a tailored cream linen suit and holding a matte black smartphone. Preserve identity, environment, pose, and lighting.
Avoid
Do not make the prompt carry every control
The biggest prompt improvement is often subtractive. Ray 3.2 rewards specific, declarative visual direction, while the controls handle preservation and freedom.
- Imperative commands like make, turn, change, or transform.
- Temporal language like throughout, over time, then, or gradually.
- Negation such as no, not, without, devoid of, or missing.
- Long mood essays that do not add concrete visual information.
- Generic adjectives that are not grounded in visible details.
Outputs
Use final settings only when the direction is ready
Ray 3.2 supports 540p, 720p, and 1080p. You may not need final resolution early. Output duration and resolution both affect cost, so early tests should stay practical.
540p
Use for internal exploration, cheap variations, and early prompt testing.
720p
Use for polished review when the direction is close enough for team or client feedback.
1080p
Use for final delivery, hero shots, broadcast, and social finals.
HDR / EXR
Use HDR for expanded dynamic range. Use EXR when the result needs professional grading or VFX finishing.
A reliable production pattern is to explore in Speed mode at 540p, then render the selected direction in Quality mode at the final resolution. Speed saves time, not cost.
Controls
Motion and Structure decide what survives
Adherence controls how strongly Ray 3.2 follows the prompt and any reference image while generating the video. Motion and Structure use a numeric scale from 1 to 9, where 9 is the strongest preservation.
Motion
Increase when timing, gestures, action, or camera movement needs to stay close to the prompt or reference image.
Structure
Increase when composition, geometry, object placement, or scene layout must remain stable.
Use Motion when movement is the thing that must survive. Use Structure when layout is the thing that must survive. Use both when an image reference, subject layout, or camera plan needs to remain highly recognizable.
Characters
Lock only the character attributes that matter
Character controls are independent locks on different aspects of a person or animal. Each enabled control tells Ray 3.2 that this attribute should survive the restyle. Toggle on what matters and leave room where the transformation needs freedom.
Faces
Locks facial identity and expression. Use it for dialogue, reactions, close-ups, and emotional beats.
Bodies
Locks silhouette, build, broad posture, and body proportions. Use it for wardrobe restyles or recognizable physical presence.
Poses
A stronger skeletal signal. Use it when choreography, hand motion, sports movement, or detailed articulation needs to stay intact.
Blocking
A lighter skeletal signal. Use it when the result should follow the general pose but can diverge into a stylized body, creature, or robot form.
Practical recipes
Dialogue close-up, style change
Faces on, Bodies on, Blocking
Locks identity and expression while preserving broad posture.
Pianist's hands, restyle the scene
Faces optional, Bodies on, Poses
Uses the stricter pose signal because subtle hand choreography is the point of the shot.
Turn a dancer into a glowing energy being
Faces off if identity changes, Bodies as needed, Blocking
Keeps sweeping choreography while giving the body design room to transform.
Crowd of distant extras
Faces off
Tiny faces rarely benefit from identity locks and may fight the transformation.
Workflow
A practical Ray 3.2 run order
- 1Choose the right input mode: T2V for building from text, I2V when an image should anchor the look or subject.
- 2Decide what the video should show, how it should move, and what visual details must stay consistent.
- 3Describe the target video in positive, concrete visual language.
- 4Use Enhance for rough exploration, and turn it off when exact wording matters.
- 5Start with a practical review setting, then escalate only the winning direction.
- 6Use Faces, Bodies, Poses, or Blocking intentionally when people or animals matter.
Start with the right input mode. Describe the end state. Choose the right adherence settings. Keep Auto on until you have a reason to change settings. That is the core Ray 3.2 workflow.
FAQ
Ray 3.2 prompt guide FAQ
What is the fastest way to improve a Ray 3.2 prompt?+
Rewrite it as the finished result. Remove commands like change, make, or transform, then add the subject, motion, style, and any reference-image detail that matters most.
Should I use negative prompts or phrases like no people?+
Use positive descriptions instead. For removal, describe the empty final scene: an empty cobblestone street at dawn is clearer than no people on the street.
When should I turn Enhance off?+
Turn Enhance off when you already have tested production wording, approved brand language, or a prompt with exact preservation cues that should be sent unchanged.
Which control should I adjust first when the result drifts?+
If motion drifts, raise Motion. If layout or object placement drifts, raise Structure. If identity or body shape drifts, use Faces, Bodies, Poses, or Blocking depending on what must stay stable.
Do I need 1080p for every test?+
No. Start lower for exploration, then render the selected direction at 1080p or Quality mode when it is ready for final review or delivery.
When are keyframes better than more prompt text?+
Use keyframes when a specific visual moment must land: a product hero frame, facial expression, end pose, style beat, or art-directed frame index.