Codex Image Generation + Seedance Image-to-Video: Chinese-Fantasy Heavenly Palace Cinematic Tutorial (2026)
# Codex Image Generation + Seedance Image-to-Video: Chinese-Fantasy Heavenly Palace Cinematic Tutorial (2026)
This tutorial uses Codex to generate a Chinese-style "Heavenly Palace" image, then uses Jimeng (Seedance 2.5) to turn it into a living cinematic scene of an ancient bridge floating above a sea of clouds, with immortal maidens walking across it. You don't need to understand how AI works — just follow the steps, paste the prompts, and you'll have a finished clip in about 15 minutes.
# TL;DR
- Three steps to a finished clip: ① Codex generates the image → ② Codex writes the image-to-video prompt from it → ③ Jimeng (Seedance 2.5) turns it into video.
- Keep the figures tiny (about 2% of the frame), leaving plenty of sky and sea of clouds, so the video has room to "move".
- The video prompt must follow your own image — don't copy someone else's word for word.
- Codex has a daily quota — if image generation fails, check your quota first.
# Preparation
- Codex (OpenAI): generates the image, and writes the image-to-video prompt based on it.
- Jimeng (Seedance 2.5): image-to-video — turns your image into an animated video.
- The three prompts provided in this article (see below).
# Step 1: Generate the Image with Codex
Paste the following prompt into Codex to generate the Heavenly Palace image:
A majestic ancient bridge extends from the foreground into the distance, curving in a magnificent arc toward a hazy amber horizon.
The bridge deck follows the arc, clean, wide and simple, with polished pearl-grey stone and lacquer that complements the massive bridge piers.
In the left foreground, a huge carved pillar of pearl-grey stone dominates; more towering pearl-grey stone columns recede along the bridge.
Right side: an open, boundless sea of clouds. Two Chinese immortal female figures, one in a long crimson robe and one in a white robe, walk slowly along the curved deck, looking tiny against the majestic architecture. The figures take up 2% of the frame height — minuscule specks of color, walking slowly forward. Their bodies stay still while their clothes and long hair catch the high-altitude breeze from one direction, flowing backward. Around them is empty space many times their height.
Warm afternoon side-lighting, creating a soft, misty atmosphere.
Distant background: majestic Chinese temple buildings faintly visible through the swirling mist; a peaceful and solemn mood, like a Chinese fairyland;
Epic Eastern aesthetics, solemn and ethereal, with dramatic scale contrast.
Expected result: an image of an ancient bridge sweeping from the foreground along an arc toward the distance, wrapped in a sea of clouds with a palace faintly visible in the background — the two maidens occupying only about 2% of the frame.
If you don't have a prompt, you can also drop a similar Chinese-style ancient-bridge picture into Codex and ask it to generate based on that reference.
# Step 2: Generate the Image-to-Video Prompt
Put the freshly generated image back into Codex and paste this instruction, so Codex looks at the image and writes a video prompt:
Describe what this image looks like in motion — give me an image-to-video prompt
Expected result: Codex returns a combined prompt covering camera movement, environment movement, and character movement. Use its output as the source of truth — it doesn't need to match this article's example word for word.
# Step 3: Generate the Video with Jimeng
Put the image and the prompt from the previous step into Jimeng, select image-to-video mode, and paste the prompt:
Recommended motion:
Camera: slow push-in (dolly in) with a slight rise, making the bridge's depth and the sea of clouds more breathtaking.
Environment: the sea of clouds slowly churns and flows, like immortal mist swirling around. The distant palace is veiled in thin mist, clouds drifting slowly across the buildings. Golden sunset light shifts, rays piercing through the cloud layers. The bridge surface faintly reflects the changing sky.
Characters: the two immortal maidens walk slowly forward along the bridge; the red-robed figure's hem and long hair sway gently in the high-altitude breeze; the white-robed figure's sleeves drift lightly. The figures stay very small in frame to emphasize the vastness of the world.
Expected result: after generating, check three things — whether the sea of clouds flows naturally, whether the figures stay minuscule, and whether the golden sunset light breaks through the clouds. If you're not satisfied, tweak the prompt and regenerate.
# Screenshots




Final result:

# Notes
- Codex quota: image generation has a daily cap. If it fails, confirm your quota first instead of retrying repeatedly.
- The source image decides everything: figures too large in the frame or an over-crowded composition will make the video look "static". When generating the image, keep plenty of open sky and sea of clouds.
- Let the prompt follow the image: each image has its own composition and lighting — base the video prompt on your own image rather than forcing someone else's.
- Jimeng queueing: generating a video usually takes some waiting — be patient, and the clip will typically come through.
# FAQ
Q: I don't have a prompt, but I still want to make a Heavenly Palace video. What should I do?
Just use the Step 1 prompt from this article and let Codex generate an ancient-bridge-over-clouds image with immortals beside it. If you're not happy with the generated image, you can also find a similar Chinese-style ancient-bridge picture, put it into Codex, and have it generate or rewrite based on that.
Q: Why is my Jimeng video stiff and static, barely moving?
Nine times out of ten it's the source image: the figures are too large in the frame, the composition is too crowded, and there isn't enough open sky or sea of clouds. When generating the image, keep the subject minuscule and leave open space — then the video has room to flow naturally.
Q: Codex generated an image but immediately said my quota is running out. What now?
Codex has a daily quota, and image generation burns through it faster than normal chat. Wait for the quota to reset the next day, or check your current plan's available usage. Don't keep retrying once the quota is exhausted.