Make AI images and videos that look like you meant it. 150+ models, one subscription. Also building https://xitter.cf/t.co/27bsHrtDph 🎬 Tag us to get featured
Joined December 2011
- Tweets26.3K
- Following1.5K
- Followers5.1K
- Likes37K
One rainy bus-stop scene, three shots, 77 credits total.
Seedance 2.5 shot the wide, Kling 3.0 gave the best acting, and Veo 3.1 Fast pushed into a 1080p close-up. All three started from one 1.5-credit still.
Full prompts in the thread below. Bookmark it.
🤖 Made with AI
Step 3, the medium. Kling 3.0 from the same still, 5 s, 27.5 credits:
"She lowers the timetable, looks down the empty road and tucks a strand of hair behind her ear, the camera drifting slowly sideways, rain beading on her coat, soft rain and a faint passing car."
Step 4, the close-up. Veo 3.1 Fast from the same still, 4 s, 1080p, 18 credits:
"Slow push in to a close up of her face as headlights sweep across it and she whispers, "Right on time." Rain on her collar, warm amber light, a distant engine rumble."
Tip: one strong still keeps her face the same across all three engines.
How it was made, step 1: a Seedream 5 still, 9:16, 2.5 credits. Exact prompt:
"Full-body vertical photo of a fictional man in his early 40s with a short gray beard, wearing a weathered olive field jacket, hiking boots and a canvas backpack, standing on a rocky mountain ridge at golden hour, looking out at the valley, clouds below, natural light, shot on a 35mm lens, realistic photo"
Dragon Ride is one photo, no prompt, and a 5 second HD clip with sound. Free accounts get 3 effects every day, no card needed.
pixeldojo.ai/effects/dragon-…
A soaked retriever on a sunset dock, and it cost zero credits.
FLUX 3 Klein for the still, MiniMax H3 Turbo for the video. Both unlimited on every Pixel Dojo plan.
🤖 Made with AI
How it was made, step 1: FLUX 3 Image Klein on Pixel Dojo, 9:16, one take, zero credits. Prompt: "A soaking wet golden retriever standing on a wooden lake dock at golden hour, mid-shake, water spraying off its fur in glittering arcs, low angle, warm backlight, calm lake and pine trees behind, photo"
Step 2: MiniMax H3 Turbo, image-to-video from that still, 9:16, 5 seconds, zero credits. Prompt: "The golden retriever shakes its whole body hard, water flinging off its fur in slow motion and sparkling in the sunset light, then it stops, looks at the camera and pants happily. Camera holds steady, low angle."
The ice cracked under the runners and the boat didn't slow down.
Every frame started as a native 4K WAN 2.7 Pro still. Seedance 2.5 added the motion and the sound of the lake breaking up.
Full prompt + settings in the thread below. Bookmark it.
🤖 Made with AI
WAN 2.7 Image on Pixel Dojo (Pro is 1.5 credits at any size, 4K included): pixeldojo.ai/wan-2.7-image
Seedance 2.5: pixeldojo.ai/seedance-2-5
Seed Audio: pixeldojo.ai/seed-audio
