The principle: don’t ask the model to invent everything
A common mistake is writing a prompt like “Monica at the Emmys” and expecting the model to preserve the same character, dress, backdrop, and red carpet atmosphere. The more details the model has to invent, the more room there is for drift.
Reference 1 · Monica
Identity, face, hair, makeup, gown, Star of David necklace, and proportions.
Reference 2 · Emmys wall
The blue wall, gold feature, carpet, logos, and the feel of a real awards ceremony photo spot.
Create a precise reference for Monica
Instead of generating a single beautiful image, we created a character reference grid. The reason is simple: in a video, the camera sees the character from different angles. If the model has only seen a front view, it starts inventing the sides, back, hair, and structure of the gown.
What the grid needs
Decisions about the look
- A champagne gold couture gown with a strong visual presence.
- A clean, structured bodice and a dramatic, voluminous ball gown skirt.
- Hair fully tied back, sleek, with a center part and a low bun.
- Polished red carpet makeup without a plastic look.
- A clearly visible Star of David necklace, identical from every angle.
- A clean white background, with no location or unnecessary elements.
A condensed character reference prompt
Create one ultra photorealistic 6 panel character reference grid of Monica. Use the provided Monica reference as her EXACT identity reference. Preserve her exact recognizable face, eye color, skin tone, age, facial structure, body proportions and identity. IMPORTANT HAIR LOCK: Monica’s hair is FULLY TIED BACK in every single panel. Precise center part. Extremely sleek hair pulled tightly away from the face into one small polished low bun. No loose hair around the face. No waves. No curls. No loose strands. No ponytail. WARDROBE: A dramatic champagne gold couture ball gown designed for a major Hollywood awards ceremony. Structured strapless corset bodice. Large luxurious full ball gown skirt. Extremely voluminous, cinematic silhouette. Long dramatic train. Modern couture, not bridal, not costume-like. JEWELRY: One elegant gold Star of David necklace. The six pointed pendant must be clearly recognizable and identical in every panel. Small diamond stud earrings only. GRID: 1. Full body front view 2. Full body three quarter view 3. Full body side profile 4. Full body back view 5. Tight beauty close up 6. Medium close up from chest upward clearly showing the Star of David necklace BACKGROUND: Pure seamless white studio background in every panel. No gray. No gradient. No props. No event environment. No text. STRICT CONSISTENCY: Same Monica. Same face. Same hair. Same bun. Same makeup. Same necklace. Same gown. No identity drift. No dress redesign. No AI aesthetic. Vertical 9:16 master image containing all six panels.
Create a reference for the Emmys location
The second reference should be a clear, clean image of the actual location, rather than something merely inspired by the carpet. Remove people in the frame and any social media interface.
What we kept
- The ribbed blue wall.
- The diagonal gold feature.
- The EMMYS lettering and the existing gold elements on the wall.
- The blue carpet.
- The real perspective of the photo spot.
What we removed
- People.
- TikTok / Instagram icons.
- Captions, comments, likes, search bars, and any other interface elements.
Bring it together in Seedance 2.5
At this stage, we uploaded just two references: Reference 1 is Monica, and Reference 2 is the Emmys wall.
Reference 1
Monica’s grid defines her identity, gown, hair, makeup, necklace, and body proportions.
Reference 2
The clean Emmys wall defines the environment, perspective, and branding already present in the image.
The directing decision
We wanted entertainment press footage, with small, natural movements. Monica is already at the photo spot. Photographers call her name; she shifts her gaze, reacts, turns, and shows the gown.
The exact prompt we used
The prompt follows a timeline. This helps the model understand both what happens and when each action should take place within the 15 seconds.
Create a 15 second vertical 9:16 ultra photorealistic live action red carpet video at the Emmy Awards. Use Reference Image 1 as Monica’s EXACT identity, wardrobe, hairstyle, makeup and jewelry reference. Use Reference Image 2 as the EXACT Emmy Awards location reference. STRICT REFERENCE LOCK Monica must remain completely identical to Reference Image 1 throughout the entire video. Preserve her exact recognizable face, eye color, skin tone, facial proportions, age, body proportions, hairstyle, makeup, champagne gold gown and Star of David necklace. Do not redesign her. Do not change her dress. Do not change her hairstyle. Do not change her jewelry. Do not alter her face between frames. The Emmy backdrop and carpet must remain visually identical to Reference Image 2. Preserve the exact blue ribbed wall, gold diagonal architectural section, blue carpet, gold Emmy statues, NBC symbols and EMMYS branding. The location must immediately read as the Emmy Awards. Do not redesign or reinterpret the location. OPENING 0:00 to 0:03 Start on a realistic handheld medium close shot of Monica from approximately mid torso upward. The framing must immediately show BOTH Monica and clearly readable EMMYS branding behind her. Her gold Star of David necklace must be clearly visible at the center of her chest from the first frame. Monica is already standing on the Emmy red carpet photo line. She is not walking toward the camera. She hears photographers calling her name from different directions. Natural camera flashes fire around her. She looks directly toward the camera with a calm confident smile. The shot must instantly communicate: MONICA AT THE EMMYS. 0:03 to 0:07 The camera naturally pulls slightly backward, revealing Monica’s full champagne gold couture gown from head to toe. She subtly changes her pose like an experienced celebrity. One hand rests naturally near her hip. She shifts her weight elegantly. She first looks toward photographers on camera left. Several camera flashes. Then she smoothly turns her gaze toward photographers on camera right. She gives a slightly bigger smile. Her movement is effortless and spontaneous, never choreographed. 0:07 to 0:11 Monica reacts naturally to another photographer calling her name. She gives a small confident laugh. She briefly raises one hand in a relaxed playful gesture as if acknowledging the photographers. Then she gives one subtle elegant blown kiss toward the press line. Nothing exaggerated. She immediately returns to a composed red carpet pose. The Star of David necklace continues to remain clearly visible whenever she faces the camera. 0:11 to 0:15 Monica slowly turns her body to approximately three quarter profile to show the silhouette and back of the gold gown. She briefly looks back over her shoulder toward the photographers with a confident smile. More realistic camera flashes. During the final two seconds she smoothly turns slightly back toward the cameras for one final pose. End while she is still posing confidently in front of the Emmy wall. CAMERA STYLE Real celebrity press footage. Handheld professional red carpet camera. Subtle natural camera movement. Occasional small reframing as the camera operator follows Monica. No perfectly stabilized commercial camera movement. No dramatic cinematic dolly movement. No slow motion. No fashion film aesthetic. It should feel like authentic footage captured by entertainment media standing directly in front of Monica at the Emmy Awards. LIGHTING Realistic red carpet event lighting. Multiple photographic flashes. Soft flattering event key lighting. The champagne gold gown naturally catches and reflects the flashes. Maintain realistic skin texture. No overexposed skin. No artificial golden glow. AUDIO No music. Authentic Emmy red carpet ambience only. Camera shutters. Photographers speaking over one another from off camera. Use natural English voices such as: “Monica! Right here!” “Monica, look this way!” “Monica, to your left!” “Beautiful!” “Monica, right here!” Voices should overlap naturally and come from different directions. Monica does not speak. She only reacts naturally through eye contact, smiling, a small laugh and body language. IMPORTANT REALISM RULES No subtitles. No captions. No added graphics. No artificial text overlays. Do not add extra logos. Do not change the existing Emmy branding from Reference Image 2. No AI aesthetic. No plastic skin. No face morphing. No wardrobe changes. No jewelry changes. No disappearing necklace. No distorted hands. No unnatural body movement. No runway walking. No exaggerated posing. No dramatic model behavior. Monica must behave like an established celebrity who is completely comfortable being photographed on a major awards red carpet. The final result should feel indistinguishable from genuine entertainment press footage from the Emmy Awards.
Mistakes to avoid
Don’t overload one prompt
Without references, the model has to invent the character, gown, and location. That is where drift begins.
Don’t use a weak character reference
One front-facing image is not enough when the character turns during the video.
Don’t leave hair loose in the grid
If some panels show a different hairstyle, the model receives conflicting instructions.
Don’t turn it into a fashion film
The goal was press footage: small movements, flashes, and glances toward photographers, rather than dramatic walking or slow motion.
Checklist before you generate
This video follows a process you can repeat.
In the OMER METHOD guides, we work at exactly this level of detail: building references that keep a character consistent, directing a scene, writing a video prompt, and bringing it all together into content that looks and feels real.
If this guide helped you, the next step is to learn the full system and use it again and again — for the red carpet or any other story you want to create.
Explore OMER METHODCharacter creation · References · Direction · Storytelling · Prompts · Workflow