Photo tips · Getting Started
Short answer: The best starting photos are sharp, well-lit, and show the subject clearly without heavy filters or obstructions. Consistent framing and visible faces or body outlines can give an image-to-video system clearer visual information, although no photo guarantees a perfect result.
Content updated: October 10, 2026
A good AI video starts with an image the model can interpret. You do not need a professional camera or a perfectly staged portrait, but a clear subject and useful framing can give the generator a better starting point. These tips apply broadly to image-based AI video tools; check the specific generator for its actual upload requirements.
Choose a photo where the subject is easy to see
Look for a photo with a clearly visible face and enough of the body in frame for the intended movement. If the subject is tiny in the distance, hidden behind another person, or partly cut off by the edge, the model has less visual information to work with. A simple background can also make the main subject easier to distinguish.
Prefer even, natural lighting
Bright, even light usually makes facial features and clothing easier to read. Very dark images, harsh shadows across the face, and strong backlighting can hide details. Natural window light or an evenly lit outdoor photo is often a practical choice. Avoid heavy filters that change facial structure or smear fine details.
Watch for blur and compression
Use the clearest version of the image you have. Screenshots of screenshots, tiny profile pictures, and heavily compressed images can lose details that matter during generation. If you are choosing between a sharp, ordinary photo and a dramatic but blurry one, start with the sharper image.
Keep framing and subject scale in mind
For dance-style motion, a crop that shows enough of the subject may be more useful than an extreme face close-up. If a workflow asks for two photos, images with reasonably similar framing and lighting can make the pair feel more coherent. Follow the tool's on-screen guidance, since different workflows may expect different kinds of images.
Quick pre-upload checklist
- The intended person or pet is easy to identify.
- The image is sharp enough to see important details.
- The subject is not mostly hidden or cut off.
- Lighting is sufficient to distinguish the face and body.
- The photo is in a file format and size accepted by the current tool.
- You have permission to upload and use the image.
What if the output still looks wrong?
Even a strong source photo cannot guarantee a clean result. AI models may alter clothing, facial details, body proportions, or movement. If a clip has obvious glitches, try a different clear photo and compare the results. Before paying for another generation, review how the service handles credits and failed attempts.
For a step-by-step workflow, read how to make a Smalltown Boy AI video with two photos. If you run into upload or generation issues, the troubleshooting guide may help.
Start with a clear image
Prepare your photo, check the live tool's supported formats and terms, and keep expectations realistic. Smalltown Boy AI videos are up to 15 seconds long, making a simple, readable idea a sensible place to start.
Try the Smalltown Boy AI generator ↗
Specific file limits and product behavior can change; confirm them on the live site.
How to compare two candidate photos
If you are unsure which image to use, compare candidates at the same display size rather than zooming in on one and viewing the other as a thumbnail. Ask whether the main subject is easy to recognize, whether the important features are visible, and whether the image has enough room around the subject for movement. A technically sharp image can still be a poor choice if the face is tiny or hidden by another object.
Do not over-edit a photo to make it look perfect. Heavy filters, artificial sharpening, or aggressive background removal can introduce artifacts of their own. Start with the clearest natural image available, and keep expectations realistic: image quality can improve the odds of a clean result, but it cannot guarantee the generator's motion or final composition.
Frequently Asked Questions
Should I use full-body or portrait photos?
Choose framing that matches the result you want and the tool’s current guidance. Keep the subject unobstructed and avoid cropping important features at the edge.
Will a higher-resolution photo always look better?
Not necessarily. Clarity, lighting, subject visibility, and the model’s input requirements also matter.
Related guides
Continue with a guide that answers the next question in the process: