Use this article to get better results from Ask Vicky, both in what you type and which model you pick for the job.
What makes a good prompt?
A few habits make a real difference to what Ask Vicky generates:
Be specific about the subject, setting, and style you want
Mention important details up front, like colors, mood, or any text you want included in the image
Ask for one clear thing at a time, rather than combining several different requests into one prompt
Use the Extra guidance field on Recipes to fine-tune a result without rewriting the whole prompt
Which image model should I use?
For high resolution, in-image text, or combining multiple reference images: use Nano Banana Pro, from 60 credits
For most everyday image generation: use Nano Banana 2, from 32 credits, it's faster and more cost-effective, and also supports editing
For text-to-image only, with no editing needed: GPT-Image-2, from 4 credits, works well, but for editing an existing image use one of the Nano Banana models instead
Which video model should I use?
For cinematic motion with synced audio and lip-sync: use Seedance 2.0, from 216 credits
For top-tier cinematic motion with native audio, especially animating a static image: use Kling Video v3 Pro, from 224 credits
For a quick, good-value result at a lower cost: use Veo 3.1 Lite, from 48 credits
Note: Each model's exact credit cost is always shown before you generate, so you can compare before choosing. See What is Model Studio and how do I use it for the full walkthrough.
What if I'm not getting the result I want?
Use Edit ad on any image in My Generations to refine a result instead of starting over. Edit ad works on generated images; videos can't be edited this way yet. Describe exactly what should change, and Ask Vicky builds on what's already there rather than generating from scratch.