A practical guide to writing prompts that get accurate, on-brand image and video results from Shoots. For an overview of what Shoots is and how to open a chat, see Shoots - Your New AI Photography Studio.
Prompting basics
Talk to Shoots like a person, not a search engine. Plain language describing the scene works better than a list of comma-separated keywords. A strong prompt usually answers:
Subject: what is in the frame
Setting: environment, surface, background
Look and feel: lighting, mood, style, colour palette
Purpose: what the asset is for
Example: "Create a clean lifestyle shot of our oat milk on a marble kitchen bench, soft morning window light, coffee accessories nearby, this image is for an advertisement."
If you don't know where to start, ask Shoots for ideas, for example "Give me three scene concepts for a summer campaign for this product."
Asking for an image vs a video
Shoots produces both formats from the same chat. By default, Shoots generates images. To create a video, include the word video in your request.
Set the aspect ratio in the prompt box before sending, or open an existing asset and ask Shoots to change it.
Format | Supported aspect ratios |
Image | 9:16, 4:5, 1:1, 16:9 |
Video | 9:16, 16:9 |
Match the aspect ratio to the placement: 9:16 for Stories and Reels, 1:1 for in-feed, 16:9 for landscape and YouTube.
Adding context with the + menu
The + menu in the prompt box is the single biggest lever for accurate, on-brand results. Use it to reference:
Products from your product catalogue or Shopify account, so Shoots uses the real product image as context
Audiences from your Brand Strategy, so the scene is built around a specific buyer
Value claims from your Brand Strategy, so messaging stays consistent
Reference images from your Albums or desktop, for dimensions, style, lighting, mood, or composition guidance
Example: "Make a UGC-style video for @Performance-Driven Parents featuring @Hydration Drink Mix, hero the @no-added-sugar value claim."
Reference images are used as guidance, not exact templates. Use one at a time when possible, and tell Shoots what to take from it: "use the lighting and mood of this image, but keep our product as-is."
Prompt formulas
For most images, this structure works well:
[Subject] + [Setting] + [Lighting] + [Style and mood]
Example: "Our skincare serum bottle on a stone bathroom shelf, soft natural daylight, calm minimalist style with warm beige tones."
For videos, separate what happens, what it looks like, and (optionally) what is heard:
[Action or shot description] + [Visual style and lighting] + [Audio note]
Example: "A slow 4-second push-in on the bottle as steam rises in the background, cinematic warm lighting, soft ambient kitchen sounds."
Helpful additions for either format:
Camera language: "eye-level," "overhead," "push in," "slow rotate"
Style cue: "photorealistic," "editorial," "cinematic"
Length (video only): shorter clips of 4 to 6 seconds follow prompts more reliably
Avoid asking for on-screen text inside videos when accuracy matters. Either leave the screen blank for an editor to add text later, or upload a screenshot of the exact text and ask Shoots to use it.
Animating an existing image
You can ask Shoots to turn an existing image into a video. Add the image in the chat using the + menu and prompt for the type of motion.
This works well when adding animation to subtle environmental motions (steam, leaves, fabric), camera moves (slow rotate, push in, pan), and lifestyle motion such as a hand reaching in or a person sipping.
Editing with a prompt
If you want to iterate or build on one particular image or video generated in Shoots, you can do so by editing:
Select the generated image or video.
Click Edit.
Describe the change, for example "make their shirt blue" or "remove the cup from the right side."
Edits create a new variant of the same scene. You can also branch the conversation at any earlier message to explore an alternate direction without losing the original thread.
If output quality stops improving
Editing the same image or video repeatedly down one chat thread can gradually reduce Shoots output quality. This shows up as results that look grainier or less sharp than earlier generations in the same chat, or as an edit that stops changing the output even after a few specific attempts to guide it. This happens because each edit builds on the accumulated context of the whole thread, so the more edits made down a single thread, the further the output can drift from the original prompt and reference material.
If a result stops improving after a few specific edit attempts, branch into a fresh chat instead of continuing to edit the same thread:
Hover over the image or video you want to keep building from.
Select Branch from the header that appears above it.
Continue prompting in the new chat. Branching carries over the context from that point, but starts a fresh thread so new edits no longer build on the degraded history.
Getting product details right
Product accuracy is the most common challenge in AI generation. Shoots can only work with what it can see, so the clearer the source you give it, the more accurate the result.
Reference the product through the + menu so Shoots uses the real product image as context, rather than guessing from text.
Use clear, well-lit product photography as the source. Front-on or three-quarter angle, neutral background, with label and packaging fully visible.
Be explicit when accuracy is critical. Add an instruction like "keep the label, shape, packaging, and colour exactly as in the reference."
Shoots fills in detail it cannot see. If part of the product is hidden in the source image, Shoots will invent that part, which often produces wrong colours or branding. Use a reference that shows the detail you want in the result.
For multi-product scenes, describe spatial relationships and rough scale. Example: "Our matcha tin sits on the left, the whisk leans against it, the bowl is in front. The tin is roughly twice the height of the bowl."
Iterate in small steps using Edit. Edit changes one thing at a time and preserves what already works. Example: "make the bottle taller relative to the bowl."
Common Shoots errors and how to fix them
If a generation in Shoots fails with an error, retry it first. Retrying resolves most Shoots errors, including errors on requests that only used images and did not ask for a video or any audio. In these cases the error is not caused by anything in your prompt, and a retry typically succeeds within a minute or two.
If retrying does not resolve the error, the error is more likely a content moderation flag from the underlying model reacting to your prompt wording. This is more common when a prompt references real, identifiable people, celebrities, or brands, or describes violent, unsafe, or otherwise sensitive imagery. To resolve a wording-related flag:
Simplify and soften the wording in your prompt.
Remove any mention of real people, celebrities, or brands.
Retry the generation.
If the error persists after a few retries, contact Support: support@cuttable.com.