A Bad Times AI video goes wrong in nine predictable ways, and every one of them is fixed by changing the photo, changing the dance or the model tier, or rendering again, never by editing a prompt. On Bad Times Dance AI you never write a prompt at all: the site builds it, so the only levers you hold are the three that matter.
Key Takeaways
- Nine problems cover almost every bad render: face drift, human fingers on pet paws, merged people, outfit changes, the wrong number of people, stiff lip-sync, timing off the beat, a blurry close-up and a blocked upload.
- Fix it in this order: a better photo first, then another dance or a higher tier (Pro or Ultra), then simply render again.
- A failed render returns its credits automatically. A render that finished but looks wrong is a normal paid render, so choose the fix before you spend again.
- Photos of celebrities, public figures and minors are blocked, and every person in the photo must be 18 or older and must have agreed.
How to Tell a Bad Render From a Normal One
Start with what the tool is doing. Bad Times Dance AI takes one photo, matches it to a reference motion clip for the dance you pick, and has a Seedance video model draw you or your pet performing the 13 or 15 second routine. The original song audio is added after the picture is rendered. Because the moves come from the reference clip, the choreography is fixed: head in hands, a wave of the hand, hand on hip, the stop palm, a walk into the camera and a finger pointed into the lens.
That means a render is only "wrong" when something that should come from your photo (a face, a coat, an outfit, a head count) does not, or when the model fumbles the movement. Anything else is the dance doing what the dance does. The close-up at the end is meant to be intense and a little theatrical. A pet raising both paws is the trend, not a glitch.
Before you change anything, watch the clip three times: once for faces and bodies, once for the movement against the beat, and once for the final close-up. Write down which of the nine problems you see. Each has a different fix, and guessing wastes credits.
The nine fixes below use only three tools. A different photo costs you nothing but a minute. Another dance or a higher model tier costs more credits. Rendering again costs the full price of that render unless the render itself failed.
Fixes for Faces and People
1. Face drift
Face drift is when the person in the video stops looking like the person in the photo, usually a few seconds in or in the final close-up. It happens most when the photo is small, shot from the side, heavily filtered, wearing sunglasses, or partly covered by hair or a hand.
The fix is a different photo. Choose one that is sharp, front-facing, evenly lit and unfiltered, with both eyes visible and the whole head in frame. A knee-up or full-body photo helps the Original template, because the first ten seconds are a wide shot and the model needs a clear sense of your proportions.
If a good photo still drifts, move up a tier. Standard runs Seedance 2.0 Mini, and Pro runs Seedance 2.0. Pro at 720p costs 20 credits against 10 for Standard at 720p, and Ultra (Seedance 2.5) costs 37. Only then render again.
2. Human fingers on pet paws
This is the best known failure in AI pet videos: the dog or cat stands up and its front paws turn into human hands with fingers. The Dog or Cat template already tells the model to keep paws, pads and claws, but it can still happen on some renders.
The fix is not to rewrite anything. First, render again on the same photo, because paws are drawn fresh each time. If they come back wrong twice, use a different photo where the pet's front legs are visible and clearly paws, ideally with the pet standing or sitting upright. Photos where the paws are hidden in fur, under a blanket or out of frame give the model nothing to copy, and it fills the gap with hands. After that, try Pro or Ultra.
3. Merged people in duo and group videos
In the Couple Duo and Friend Group dances, two bodies sometimes fuse at the shoulder, or a face is shared between two people. This comes from the photo: people hugging, overlapping, leaning into each other or standing so close that the edge between them is unclear.
Use a different photo with everyone side by side and facing the camera, nobody half hidden. Different-coloured tops help the model keep two people apart. Keep a group to three or four people; five usually works, but beyond that each face is small and the model starts to blend them. For exactly two people, choose Couple Duo, not Friend Group, because each face gets more detail.
4. Outfit changes
The outfit in the video should match the outfit in the photo. When a jacket becomes a t-shirt or a hat vanishes, the model has usually been given too little to work from. A cropped selfie shows only a collar, so everything below the frame is invented.
The fix is a photo that shows the clothes you want, waist-up at least and full-body if you can. Plain, solid clothing is easier to carry over than busy patterns. If you are using the Solo template with a portrait, expect jeans or shoes below the frame to differ from what you wore. Pet outfits follow the same rule: a sweater or bandana in the photo stays on, and one barely visible may not.
5. Wrong number of people
You uploaded two people and got three, or a group of four came out as five. Stray people in the background are the usual cause. A passer-by, a figure in a mirror or a picture on the wall can be treated as another performer. This is also a safety point: Bad Times Dance AI requires every person in the photo to be 18 or older and to have agreed, so a crowd photo with strangers in it is not allowed in the first place.
The fix is a photo with only the people who should dance, against a plain background. Crop out strangers before you upload. If the count is still wrong, pick the dance whose cast fits: Solo Spotlight or the Original for one person, Couple Duo for two, Friend Group for three or four, Pet Duo for two pets.
Fixes for Movement and Sound
6. Stiff lip-sync
The song is part of the trend, and the final close-up asks the face to mouth the words like an argument. When the mouth barely moves, or moves out of time with the voice, the video feels dead. Lower tiers have less capacity for fine facial motion.
Two fixes help. A photo with a relaxed, neutral mouth gives the model room to animate; a wide grin or an open-mouth shot can leave less to move. Then move up the model ladder. Pro (Seedance 2.0) and Ultra (Seedance 2.5) draw the close-up with more detail. At 720p, Pro is 20 credits and Ultra is 37. If you want to compare before committing, one render on each tier from the same photo tells you faster than any guesswork.
7. Timing off the beat
Each dance is matched to its own audio, and the moves are copied from a reference clip, so the beats should land: head in hands at the start, the stop palm around the middle, the walk-in around the 10 second mark, and the finger in the lens through the final seconds of the 15 second dances. If a move arrives late, or the walk-in feels rushed, the model has drifted from the reference.
Do not try to fix this by trimming the file. Choose another dance, since a different template uses a different motion clip, or render the same photo on a higher tier. If the problem is only in one place, a second render of the same dance often lands better, because each render is a fresh draw.
8. Blurry close-up
The last four seconds of the 15 second dances are a tight close-up, the part viewers watch most closely. A soft face in those seconds is nearly always a resolution or photo problem. Low resolution (480p) will always look softer than 720p, and a small, blurry or heavily compressed source photo cannot produce a crisp face.
First, upload the largest, sharpest photo you have, with the face large in the frame. Second, raise the resolution. For Standard, 480p is 5 credits and 720p is 10. Pro offers 1080p at 50 credits, and Ultra 1080p is 93, so reserve those for a video you really want to post. Moving from 480p to 720p is the cheapest improvement and often enough.
When the Upload Is Blocked
9. Blocked upload: celebrities, public figures and minors
Sometimes the problem is that nothing renders at all. Bad Times Dance AI does not allow photos of celebrities, public figures or minors, and such uploads are blocked or removed. The minimum age is 18 for everyone in the photo, and each person must have agreed to the video. This is part of why the site refuses to make a baby version of the trend.
There is no workaround and none is offered. The fix is a different photo: yourself, friends who are adults and agree, or your own pets. A photo with a recognizable person in the background can trigger a block too, so crop the frame to the people who belong in it. For the rules in full, read the site's acceptable use page at /acceptable-use.
Photos are not kept long: uploaded photos are deleted within 24 hours, and the finished video stays downloadable for 24 hours, so download your file as soon as it finishes.
The Fix Ladder and What a Retry Costs
Put the nine fixes in one order and you will rarely waste a render.
- Change the photo. Free, and it solves face drift, merged people, outfit changes, the wrong head count, blurry close-ups and blocked uploads.
- Change the dance. Free to choose, and it helps with timing and with casts that do not fit the template.
- Raise the tier or resolution. Standard 720p is 10 credits, Fast 17, Pro 20 and Ultra 37. Use this for stiff lip-sync and persistent drift.
- Render again. Because each render is a fresh draw, this alone clears many paw and timing glitches.
The money is simple. The Single pack is $6.99 for 10 credits, which pays for one Standard 720p video. The Duo Pack is $29 for 50 credits: five Standard 720p videos, or two Pro 720p renders with 10 credits left. The Crew Pack is $99 for 250 credits.
Two rules about money are worth knowing. If a render fails outright, the credits are returned automatically. If it finishes but you dislike it, the credits are spent, and credit packs are not refundable, so a few seconds spent on the photo before you press the button is the best saving you have.
For the broader picture of how a video is made, see the how-to guide. For pets, the pet photo guide goes deeper, and once the video looks right, posting it to TikTok, Reels and Shorts is next.
FAQ
Why does my Bad Times AI video look different from my photo?
Bad Times Dance AI draws face, hair and outfit only from your photo, so a small, side-on, filtered or partly covered picture produces drift. Upload a sharp, front-facing, evenly lit photo with the whole head in frame. If a good photo still drifts, try the Pro or Ultra tier, then render again.
How do I fix human hands on my dog or cat in the video?
Render again first, since paws are drawn fresh each time. If human fingers keep appearing, upload a photo where the pet's front paws are clearly visible, then try Pro or Ultra. Bad Times Dance AI's Dog or Cat template already tells the model to keep paws, pads and claws, and failed renders return credits.
Why are two people merged in my Bad Times AI couple video?
People who overlap, hug or stand shoulder to shoulder in the photo can fuse during the dance. Upload a photo from Bad Times Dance AI's Couple Duo flow where both people stand side by side, face the camera and wear different coloured tops. Keep groups to three or four people.
Why was my Bad Times AI photo upload blocked?
Bad Times Dance AI blocks photos of celebrities, public figures and minors, and requires every person in the photo to be 18 or older and to have agreed. Use a photo of yourself, adult friends who consent, or your own pets. Uploaded photos are deleted within 24 hours.
Do I get my credits back if a Bad Times AI render fails?
Yes. A render that fails outright returns its credits automatically on Bad Times Dance AI. A render that completes but looks wrong still counts as a normal paid render, and credit packs are not refundable, so check the photo and tier before you press Make my video.
Sources
- Songfacts: Bad Times by Imael Angel
- AI Musicpreneur: five AI acts on the iTunes Top 20
- Official Charts: Imael Angel
- AI Miracle: Bad Times AI trend guide