Video generation costs more than text — iterate cheaply first.
Use this workflow to reduce the number of paid video generations it takes to get the result you want.
Draft your video prompt in a text chat first. Refining wording with a text AI is nearly free; video renders are not.
Give the text AI full context: which mode you're using (text-to-video or image-to-video), exactly what went wrong last time, and what you want. Then ask it to write the video prompt using the formula: Subject + Motion + Environment + Camera + Style + Audio.
Ask for 2-3 video prompt variants and test them — small wording changes often produce dramatically different videos.
Test at short duration / lower quality first if the video model offers quality settings, then re-run the winner at full quality.
Check the on-screen credit estimate before running a generation to understand the potential cost. Shorter or lower-quality runs typically consume fewer credits, making them a cost-effective choice for initial testing.
Keep a prompt log of what worked and what failed.
Review your credit usage history via Settings → Usage to correlate credit spend with prompt experiments and identify cost-saving opportunities.
Generate videos directly in Studio → Video instead of through the chat assistant to see and confirm the model and its associated credit cost upfront.
Opt for standard models unless a premium model is specifically required, as standard models are generally more cost-effective.