You have no footage to edit and no photograph to upload. You do have one idea: a quiet weekend coffee clip. That is enough to prepare a text-to-video attempt. The first job is to tell the generator what the cup, table and camera should be doing.
This walkthrough follows that one idea into a complete prompt, then adapts it for a rainy window and a sunny breakfast table. The prompts are ready to copy. The examples describe intended scenes and do not report a generation test.
Start with the coffee, without uploading an image
UGC Maker AI (https://ugcmaker.org/free-ugc-video-generator/) https://ugcmaker.org/is one option for trying the coffee prompt in this guide.
Use its dedicated free UGC video page for this exercise. The documented browser workflow allows a written scene without image references. Leave both Start Frame and End Frame empty, and put the prompt in the text field. You do not need to create two pictures before trying this route.
The free page describes access without signup and displays “powered by MiniMax H3”. Its separate free MiniMax H3 page also uses the label “unlimited”. Those are the pages’ access descriptions, not a promise about waiting time, resolution, clip length or access to every paid model. The professional tools and external API links are separate offerings.
For the first coffee composition, choose 16:9 so the table and window have room around the cup. The documented layout choices also include 9:16 and 1:1, but you can leave those for a later version rather than changing the framing and scene together.
Turn “quiet weekend coffee” into this first prompt
The short idea leaves almost everything undecided. “A plain ivory ceramic cup” gives the scene a subject. “On a wooden table beside a window” places it. “Soft morning light” describes its appearance. A slow camera push provides the intended change while the objects remain still.
Copy this complete version:
Create a quiet weekend coffee scene. Show exactly one plain ivory ceramic cup filled with dark coffee, with one simple handle pointing right, on one matching ivory saucer. Place it on a wooden table beside a window in soft morning light. Make a slow, restrained camera push towards the cup. Keep the cup, saucer and table still. No hands, pouring, extra cups, lettering, logos, camera orbit or scene cuts.
You can already see what a later result should be checked against: one cup, one saucer, a window-side table and a modest push towards the subject. “Quiet” no longer has to carry the whole instruction.
These words specify an intention. They do not guarantee that the model will preserve the handle or execute a precise camera path. This is also a generated coffee concept, not a faithful record of a particular café or a branded mug you own.
Make it a rainy window-side coffee
Keep the cup and camera instruction. Replace “soft morning light” with “soft, cool overcast daylight, with rain droplets visible on the outside of the window”. That is the main setting change; you do not need to add a person, a moving curtain or a pouring action to make it rainy.
Here is the full rainy version, so you do not have to assemble it from fragments:
Create a quiet rainy-day coffee scene. Show exactly one plain ivory ceramic cup filled with dark coffee, with one simple handle pointing right, on one matching ivory saucer. Place it on a wooden table beside a window in soft, cool overcast daylight, with rain droplets visible on the outside of the window. Make a slow, restrained camera push towards the cup. Keep the cup, saucer and table still. No hands, pouring, extra cups, water on the table, lettering, logos, camera orbit or scene cuts.
The window supplies the weather context. The cup is still the subject. If you decide that running droplets should become the main movement, make that a separate attempt with a fixed camera, rather than adding every kind of motion to this one prompt.
Or move to a sunny breakfast table
For a brighter version, change the lighting phrase to “warm morning sunlight and gentle window shadows”. You can try that lighting change on its own first.
If you also want an explicit breakfast prop, add “one small side plate holding one slice of plain toast, to the right of the cup”. The complete version below includes that additional choice. It is another designed input, not a controlled comparison of rendered videos.
Create a quiet sunny breakfast coffee scene. Show exactly one plain ivory ceramic cup filled with dark coffee, with one simple handle pointing right, on one matching ivory saucer. Place it on a wooden table beside a window in warm morning sunlight with gentle window shadows. Add one small side plate holding one slice of plain toast to the right of the cup, separated from the saucer. Make a slow, restrained camera push towards the cup. Keep all table objects still. No hands, pouring, extra cups, extra food, lettering, logos, camera orbit or scene cuts.
The reusable part is still the same: cup, place, light and one camera intention. When replacing the subject with your own idea, change the concrete nouns and keep the request small enough to inspect. A plain cup is an easier description to evaluate than “a perfect café breakfast with everything”.
Generate, watch, then decide what to keep
Paste one complete prompt, confirm that both image fields are blank and choose the layout. Use Generate. If a clip completes, preview the whole candidate before using Download to save it. Do not treat a static preview or an unfinished request as a completed video.
For the coffee example, look at the cup rim, handle and saucer throughout the candidate. Do they stay recognisable, or change shape as the camera approaches? Count the cups. Check whether a second mug, an unwanted hand or unrelated food has appeared. For the rainy version, inspect whether the weather remains outside the window rather than turning into a spill on the table.
Listen as well as look. The documented service includes generated audio, but this guide does not promise a controllable soundtrack or a particular sound. Review anything produced before including it. Add exact captions, replace or remove unsuitable audio, and assemble separate clips in an external editor using material you may use.
Revise the part that caused the problem
If an unwanted object appears, avoid replacing the whole prompt with a longer list of adjectives. Identify the mismatch and revise its instruction.
If two cups appear: keep the singular subject explicit: “exactly one cup and one saucer; no other drink vessels”. Remove broader phrases such as “a complete café table” if you added them. Try that change before adding more props.
If the handle bends: simplify the cup description and make a separate fixed-camera attempt. Replace the camera sentence with “Keep the camera locked on the cup” rather than requesting a push and a fixed camera together. Inspect the next result; simpler wording is an option to try, not a guaranteed repair.
If the camera circles the table: remove conflicting movement words you may have added. Keep just “a slow, restrained camera push towards the cup” and the instruction against an orbit. Do not combine push, pan, rotation and zoom requests in the same small attempt.
Keep a copy of the prompt you used so you know which instruction belongs to which candidate. Describe a result only after you have actually watched it. The three examples here are starting inputs, not a ranking of tested outputs.
Begin with the plain weekend cup. Once you have something worth keeping, the same small prompt can take a rainy-day or sunny-breakfast direction without needing a stock-footage search first.