Grok Imagine (from xAI) is Soup.video’s go-to engine for fast, stylized AI video and quick social iteration. This tutorial covers both Grok Imagine (text/image-to-video) and Grok Imagine 1.5 (image-to-video flagship).

• Studio: Grok Imagine
• Remix / I2V: Grok Imagine 1.5

Which Grok model should you use?

Model	Best for	Durations	Resolutions
Grok Imagine	Text-to-video, fast concepts, fun/spicy modes	6–30s	480p, 720p
Grok Imagine 1.5	High-quality image-to-video from one still	5–15s	480p, 720p, 1080p

---

Grok Imagine pricing on Soup.video

Credits are ~$0.01 each. Sample Grok Imagine matrix:

Resolution	6s	10s	15s	30s
480p	~$0.10	~$0.15	~$0.25	~$0.75
720p	~$0.15	~$0.25	~$0.30	~$1.00

Grok Imagine 1.5 is per-second:

Resolution	Approx. per second
480p	~$0.10/s
720p	~$0.15/s
1080p	~$0.25/s

Top up on Pricing — cheaper than locking a separate xAI or aggregator subscription.

---

Tutorial A: Text-to-video with Grok Imagine

1. Open /video/grok-imagine

2. Write a motion-forward prompt

Grok responds well to clear verbs and vibe words.

Neon scooter races through rainy Bangkok alleys, handheld tracking shot, magenta reflections on wet asphalt, high energy, cinematic, 9:16

3. Set controls

• Aspect ratio: 9:16 for Reels/TikTok
• Mode: normal (default), fun, or spicy
• Duration: start at 10s
• Resolution: 720p for shareable drafts

4. Generate and iterate

Change only the camera line or only the lighting line between takes to learn what Grok locks onto.

---

Tutorial B: Image-to-video with Grok Imagine 1.5

1. Open /video/grok-imagine-1.5

2. Upload a clean anchor image

Portraits and product shots with a clear subject work best. Soft backgrounds beat clutter.

3. Describe the motion — not a brand-new scene

Good:

Subtle smile, hair moves in a soft breeze, slow push-in, natural daylight

Bad:

Turn this into a dragon flying over a volcano with explosions

4. Pick resolution/duration

Draft at 480p / 5s, then finalize at 1080p.

---

Prompt cheatsheet for Grok Imagine

Add: subject action, camera move, lighting, mood, aspect intent
Avoid: long lore paragraphs, conflicting camera directions, tiny unreadable text requests

More patterns: AI video prompting guide.

---

Generate Grok via API

curl -X POST https://soup.video/api/v1/generations \
  -H "Authorization: Bearer sk_soup_your_key_here" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "grok-imagine",
    "inputs": {
      "prompt": "A skateboarder ollies over a puddle at golden hour, slow motion, 9:16",
      "aspect_ratio": "9:16",
      "mode": "normal",
      "duration": "10",
      "resolution": "720p"
    }
  }'

Docs: /docs/generate.

---

Start cooking: Grok Imagine studio · Grok Imagine 1.5