Happy Horse 1.0 Text to Video: pick the row, not the brand (3s, 4s, 5s, 1080p, 28cr)
Happy Horse 1.0 Text to Video is one catalog generate. Use it because this job matches this row.
Happy Horse 1.0 Text to Video is one catalog generate. Use it because this job matches this row.
Happy Horse 1.0 Text to Video is Alibaba's text-to-video line: prompt in, clip out, no image required. Output tops at 1080p. Durations run 3s through 15s in one-second steps. Headline credits: 28. The AI video generator is the button. The brand name is not the brief.
The duration ladder is the product
The row lists 3s, 4s, 5s, 6s, 7s, 8s, 9s, 10s, 11s, 12s, 13s, 14s, 15s. That is not a marketing range. That is the clip you will get. A hook that needs two seconds is not this model — the floor is three. A scene that needs sixteen is not this model either. You pick a length that exists, or you cut.
Title shorthand says 3s, 4s, 5s because those are the lengths people actually ship on a feed. The catalog still goes to 15s. Do not slide to 15 because the control goes there. Slide to 15 because the action needs twelve and you want a tail.
1080p is the ceiling (max_output_resolution). Qualities listed: 720p and 1080p. Aspects: 16:9, 9:16, 1:1, 4:3, 3:4. Pick the ratio for the account, not a square you will crop twice.
Foley is in the description. Write it.
The audio flag on the row is empty. The description is not: "expressive videos from text prompts with native synchronized audio, Foley sound effects, and multilingual lip-sync — supporting up to 1080p output and durations from 3 to 15 seconds." Features include native_audio, foley, multilingual_lipsync.
Treat that as a warning, not a gift. If you do not name the sound — footsteps, room tone, the language of the mouth — you still may get a soundtrack you did not choose. Native audio is not a silent preview with a mute toggle. Write the bed. Write the language. Or plan a replace pass.
A talking close-up in a language you did not prompt is not "expressive." It is a recut.
28 credits is why you opened this row
Credits on the snapshot: 28. Not a reason to generate three lengths "to compare." One length, one prompt that matches text-to-video, one 1080p take. If the job is a locked product still, this is the wrong family — no image is required here because no image is used as the contract.
The Happy Horse roster has 1.0 and 1.1 lines. This page is 1.0 text-to-video only. If you needed Wan, you wanted Wan's roster. Same parent company is not the same row.
When the slate is "expressive 3-to-15-second 1080p from text, sound in the prompt," this is the generate. When the slate is a plate, a reference stack, or a 720p experiment, leave. The ranked list for the category is best text-to-video model. Use it to pick. Then run the row you picked.
FAQ
Why not just pick "Happy Horse" and generate?
Because Happy Horse is a family. This row is text-to-video, 3–15s, 1080p, 28 credits, no still required. A different Happy Horse row is a different job. Match the job to the line you are on.
Does this model require a start image?
No. Requires image is false. Categories: text-to-video. If the label on the still is the whole brief, do not start here.
Will I get audio even if I forget to write it?
The catalog description claims native synchronized audio, Foley, and multilingual lip-sync. The audio field itself is empty. Prompt the sound. Do not discover a language at export.
Is 15 seconds the default I should use?
No. The ladder starts at 3s. Use the shortest length that holds the action. 28 credits is the row price, not a voucher for the longest clip.