The best part is that you do not need complicated traditional VFX software to create this type of video. With a clear face photo, a few PNG references and a detailed AI video prompt, you can create a similar cinematic sequence using AI tools.
In this article, we will first create a consistent character image with ChatGPT and then use that image along with object and location references to generate the final AI collision video.
🎬 Watch the Video First
Step 1: First Create Your Character Image With Chatgpt
First, we need to create a clear character reference image. This image will help the video AI maintain the character's face, hairstyle and outfit throughout the video.
1. Copy Image Prompt From the Copy Button 👇
Click the button below to copy the complete image-generation prompt.
Create a photorealistic full-body character reference sheet on a seamless white background with soft, even studio lighting. Use my uploaded selfie as the identity reference. Preserve the same facial structure, facial features, skin tone, hairstyle, hair texture and natural appearance throughout every panel. Keep the character identity highly consistent. The character is a young stylish man wearing a premium dark navy blue ribbed half-sleeve polo shirt with a short zip collar, black relaxed-fit trousers and clean white low-top sneakers. Add a simple premium wristwatch. The outfit should remain exactly consistent in every view. Use stylish black rectangular sunglasses with dark lenses as part of the character's appearance. The character should also hold a bright orange smartphone naturally in one hand. Create four panels arranged horizontally: 1. Extreme close-up portrait of the face, showing the facial details clearly. 2. Full-body side profile view from head to toe. 3. Front full-body view showing the complete outfit clearly. 4. Full-body back view from head to toe. Keep the same face, hairstyle, sunglasses, outfit, body proportions, shoes and accessories in all four panels. No costume, no suit, no outfit changes, no extra people, no duplicate character. Clean white background, realistic skin texture, natural proportions, professional studio photography, sharp details, photorealistic quality, consistent identity.
2. Open ChatGPT
After copying the prompt, open ChatGPT and start a new chat.
3. Upload Your Clear Face Photo
Upload a clear photo in which your face is properly visible. Avoid heavily filtered, blurry or extremely dark photos. Then paste the copied image prompt and send it.
4. Generate & Download the Image
Wait for ChatGPT to generate the character reference sheet. Check the face, hairstyle, outfit and different views. If everything looks correct, download the generated image and keep it safely for the next step.
Step 2: Download All Required PNG References
Before creating the video, keep the following PNG reference images ready:
- Character PNG: The image created with ChatGPT.
- Location PNG: Your selected urban environment.
- Sunglasses PNG: The sunglasses you want to appear in the video.
- Mobile PNG: Your phone reference.
- Coffee PNG: Your cold coffee reference.
Download The Given Png From Download Button
Step 3: Copy the AI Video Prompt
Now copy the video prompt below. This prompt is designed to create the walking, collision, bullet-time and cinematic slow-motion effect.
Vertical 9:16 format, continuous cinematic shot in one single take, photorealistic real-life imagery, realistic physics, strong spatial continuity, premium fashion-film cinematography, natural lighting, realistic materials, shallow depth of field, and a practical high-speed photography feel. 0.0–2.5 s: The video begins at the feet of the main character from <<image_1>>, walking naturally through an urban environment inspired by <<image_2>>. The camera smoothly tilts upward along the character's body and transitions into a front close-up. The character wears <<image_3>> naturally and holds <<image_4>> against one ear as if having a phone conversation, while naturally holding <<image_5>> in the other hand. 2.5–4.0 s: A rushed person approaches head-on carrying folders and loose papers and accidentally bumps into the main character. Papers scatter dramatically in different directions. The main character reacts with genuine surprise, loses balance, and begins falling backward. <<image_3>>, <<image_4>>, and <<image_5>> visibly separate from the character because of the collision, following physically believable trajectories. 4.0–5.0 s: Time rapidly slows down into a dramatic bullet-time effect. The character continues falling in extreme slow motion while the scattered papers and released objects remain suspended around them. 5.0–6.2 s: The camera moves closer toward <<image_3>>. The sunglasses remain nearly frozen in their natural position in space, with realistic reflections and shallow depth of field. The falling character remains smoothly visible in the background. 6.2–7.4 s: The camera smoothly pans across the same frozen scene toward <<image_4>>, which is visibly separated from the character. Preserve the phone's exact appearance and recognizable details. Keep it physically suspended in the same position with realistic lighting and reflections. 7.4–8.8 s: The camera continues toward <<image_5>>, suspended after being released from the character's hand. Show the cold coffee naturally reacting to the collision and inertia. Any liquid movement, ice, straw, lid, and cup motion must follow realistic gravity, inertia, and material behavior. Preserve the exact appearance of the reference object. 8.8–10.2 s: The camera pulls back to reveal the complete frozen scene once again: the character falling backward, papers floating throughout the environment, and all three reference objects occupying their natural positions in space. 10.2–11.2 s: Time abruptly snaps back to normal speed. Gravity and momentum resume simultaneously. The character, papers, sunglasses, phone, and cold coffee fall naturally from their current positions toward the pavement. 11.2–13.0 s: After hitting the pavement, the main character briefly remains on the ground. Nearby pedestrians react naturally with surprise and look toward the incident, creating a realistic aftermath moment. Throughout the entire sequence: Maintain the exact appearance, identity, proportions, textures, colors, and recognizable details of every reference image. All three objects must originate from the character and visibly separate during the collision. Never make objects appear spontaneously. No duplicates. No teleportation. No object regeneration. No objects flying directly into the camera lens. Maintain consistent character identity, hairstyle, clothing, sunglasses, phone, cold coffee, environment, lighting, scale, and spatial positioning throughout the entire shot. The bullet-time objects must remain physically anchored in space while the camera travels around them for cinematic close-ups. Premium cinematic fashion-film aesthetic, photorealistic human movement, realistic collision physics, natural shadows, realistic reflections, subtle motion blur, shallow depth of field, high-end commercial cinematography, practical high-speed-camera look.
Step 4: Open an AI Video Generator
You can use an AI video-generation platform such as Flow or Seedance AI.
🎬 Open Flow AI 🎥 Open Seedance AI
Step 5: Create a New Project
Login or sign up if required. After opening the video tool, click New Project and open the image/reference upload section.
Step 6: Upload Your Images
Upload the generated character image along with your downloaded PNG references.
- Character image generated with ChatGPT
- Location PNG
- Sunglasses PNG
- Mobile PNG
- Cold Coffee PNG
Keep the image order consistent with the references used in the video prompt.
Step 7: Paste the Video Prompt
Paste the copied video prompt into the AI video prompt box. Make sure all reference images are uploaded correctly before generating the video.
Step 8: Choose Ratio & Quality
Select 9:16 vertical ratio for Shorts, Reels and other mobile-friendly platforms. Then select the best available quality according to your tool and generation limits.
Step 9: Generate the Video
Click the Generate / Send button and wait for the AI to create your cinematic collision video. Generation time may vary depending on the platform and selected quality.
Step 10: Download Your Final Video
Once the video is generated, preview the complete result. If the character, objects, collision and bullet-time effect look good, click the Download button and save the final video.
Important Tips for Better Results
- Use a clear and high-quality face photo.
- Keep all object references visually consistent.
- Use the same character image throughout the workflow.
- Check that the phone, sunglasses and coffee do not duplicate.
- If the first generation is not perfect, improve the references and regenerate.
Disclaimer
This tutorial is created for educational and creative purposes. AI-generated images and videos may contain visual inconsistencies, especially with faces, hands, objects, text and realistic physics. Results can vary depending on the AI tool, model, prompt and reference images. Always follow the terms and policies of the AI platform you use.
Conclusion
Creating a Trending Collision Video With AI becomes much easier when you follow a proper reference-based workflow. Start by creating your character image with ChatGPT, prepare the required PNG references, then upload everything into an AI video generator and use the detailed video prompt.
The key is not just the prompt but also good reference images, consistent character details and correct object mapping. With a few attempts and small prompt improvements, you can create cinematic collision, bullet-time and fashion-style AI videos suitable for short-form content.
