From an idea to a result you can use.
Choose a model, write a clear brief, control your credits and review your work before sharing it.
Last updated:
Start with the task.
Define the result you need before choosing a model: an image, an edit, a moving shot, a voice, a 3D asset or a text response. Use the model library to compare the tasks, supported inputs and settings that are currently available.
A model name alone does not tell you whether it fits your brief. Check the creation form for required files, output options and limits. Model availability can change; use the current catalog and an accepted price quote when you are ready to create.
| Your goal | A useful starting point | Check before creating |
|---|---|---|
| A new image | Text-to-image | Composition, aspect ratio and required text |
| Change an existing image | Image-to-image or editing | Supported reference inputs and details to preserve |
| Animate a still | Image-to-video | Starting frame, duration and motion controls |
| Speech from a script | Text-to-speech | Voice options, pronunciation and language |
| A mesh for a project | Text-to-3D or image-to-3D | File format and suitability for your target software |
| Writing or analysis | A supported language task | Context limits, output length and factual accuracy |
Give the model a clear brief.
Describe the subject, what it should do, the setting and the desired finish. Include the few details that make the result useful, then state what must stay consistent. Avoid instructions that ask for conflicting styles, camera moves or compositions.
For an image, specify framing, lighting and material. For a video, describe one main action and one camera movement. For speech, prepare the exact script and pronunciation notes. For writing, explain the audience, purpose and desired structure.
- Image brief: A small amber glass bottle on warm stone, centered with room for a headline, soft window light, realistic reflections, portrait composition.
- Motion brief: A slow camera move toward the bottle while a soft reflection crosses the glass; preserve the bottle shape and label.
- Revise one important variable at a time. Keep a copy of the prompt, model and settings so that you can compare the next result fairly.
Prepare references you have the right to use.
Use clear reference files with the subject visible and the intended crop already considered. An ambiguous or heavily compressed reference can make it harder to judge whether the result preserves the details you need.
Confirm that you may upload and use each image, recording, logo or other asset. Get permission before using another person’s likeness or voice. Avoid including private information that is unnecessary for the creative task.
Plan the iterations as well as the final result.
Decide how many attempts you can afford before starting. Where a model supports it, test composition or motion with a shorter duration, a smaller output or fewer variations. A successful test is a useful direction, not a guarantee that a later generation will match exactly.
Review the Renderly estimate for your exact settings before every creation. Through the API, a quote does not spend credits; review its maximum credit amount before submitting the job. Changing the model, files or settings requires a matching quote.
- Compare a small set of candidates using the same brief and the same review criteria.
- Spend the next attempt on a specific improvement: framing, motion, pronunciation or a missing detail.
- For an uncertain API submission, retry the original request ID and body. A new request ID can create a second billable job.
Inspect the output where it will be used.
Review the full output, not only its thumbnail or first frame. Check images at the intended display size, watch video through to the end and listen to speech with the script in front of you. Open 3D assets in your target software before committing to them.
Look for altered product labels, incorrect text, anatomy errors, motion discontinuities, unwanted speech and details that change between frames. Verify factual claims and code independently. Generation is a draft step; your review determines whether the result is ready.
- Confirm dimensions, crop, duration and format against the delivery brief.
- Check that a person, product or brand is represented accurately and with the necessary permission.
- Review the selected model’s license and usage conditions before commercial publication. AI generation does not remove rights in uploaded assets or guarantee exclusivity.
- Add an appropriate disclosure when viewers could otherwise mistake generated material for an authentic recording or event.
Keep the useful parts of your process.
Save your final files, approved brief, model, settings and review notes together. Through the API, completed jobs provide artifact IDs; request a temporary download URL and save the file before that URL expires. Request a fresh URL when you need another download.
If a result or request needs attention, include its job ID, the approximate time and a concise description when contacting us. Leave passwords, API keys and unnecessary personal information out of the report.