Free AI Video Course for Beginners: Photo to Shorts in 7 Lessons

Most people try AI video the expensive way. They type a long prompt into a text-to-video model, get a face they didn't ask for in a room they didn't pick, retake it and get someone else entirely. This course goes the other way: start from one still photo, lock the look, then add motion. It costs less, keeps your character identical from clip to clip, and it's how every workflow in the Spicy Studio playbook is built.

Over seven lessons that photo becomes a set of vertical clips for YouTube Shorts, Instagram Reels and TikTok: a small animated moment, a talking clip in your own voice, a filmed movement remade with your character, and a scripted sketch with AI voices. Each hands-on lesson ends with something you can post, not a quiz.

How it connects to the tools. Each lesson links to the matching page of the Spicy Studio playbook, where most prompts sit next to a button that opens the right lab with the model selected and the prompt filled in.

The running example. The prompts follow one fictional character, Ines, 31, who runs a small plant shop and posts quick tips and jokes. Swap in your own subject wherever she appears.

Free, no account: this page and all seven lessons.

Free account: the playbook pages open once you sign in with a free Spicy Studio account. New accounts get 25 credits on sign-up, with no card required.

What 25 credits cover: one finished first frame (Lesson 2), spent as 5 credits for a face reference (GPT Image 2 at 1K, low quality), 10 for the face-replaced photo (1K, medium) and 10 for an enhance pass on Nano Banana 2 at 2K. Not a video clip, though: the cheapest one in this course, 5 seconds of Seedance 2.0 at 480p, costs 35 credits.

Paid credits: one-time top-ups start at €2 for 90 credits and never expire. Spicy Basic is €10 a month for 600 credits, which reset each billing cycle. Every plan includes every model.

Pro: two playbook workflows, complex motion (Lesson 5) and sketches (Lesson 6), need an active Spicy Pro plan (€25 a month, 1,800 credits) or higher. The models behind them work on any plan; Pro unlocks the tested steps and prompts.

Not included: a certificate.

Goal: a one-line plan for your first clip. Cost: nothing.

The playbook splits every video into five stages: Idea → First frame → Motion → Edit → Post. It's tempting to jump straight to Motion. Start at Idea, because the idea decides the tool.

Borrow a format, not a video. Scroll Shorts, Reels and TikTok for a structure that keeps returning in your niche: a reveal, a reaction, a quick tip to camera, a street interview, a POV. Pick one example and take it apart: why you didn't swipe away, when the payoff arrives, how short it is. Then write your own version.

Pick the route by what has to happen on screen: - One small action, like a glance, a reveal or a laugh: animate (Lesson 3). - Your exact words in your own voice: lip-sync (Lesson 4). - A movement you can film yourself: motion control (Lesson 5). - A moving camera, a spin or fast action: complex motion (Lesson 5, Pro). - A scripted exchange with AI voices: sketch (Lesson 6, Pro).

Shorts, Reels and TikTok are vertical feeds, so every frame here is made at 9:16.

Ines's plan: “Reveal: Ines lifts a cactus out of a delivery box and grins at the camera. Route: animate.” That line is your Lesson 1 output. The playbook's overview page shows a result for each route, and the model list explains what each model is for.

Goal: one 9:16 still that's sharp enough to animate. Cost: 20–25 credits. Playbook page: first-frame workflow.

Every video model moves the image you hand it, and none can repair a soft face or a cropped hand. This is the lesson to get right.

1. Lock one face. If the character is you, use a clear, front-facing photo of yourself. Otherwise generate one portrait in the Image Lab and keep it for good. For Ines: “Vertical 9:16 phone photo, head-and-shoulders portrait of a 31-year-old woman with a dark chin-length bob, light freckles and a small silver nose ring, plain olive sweater, looking into the lens, soft daylight, white wall, natural skin texture.” A low-quality version on GPT Image 2 at 1K is 5 credits. Attach this exact file to every generation; switch to another and viewers meet a different person.

2. Find a base photo. Take a real photo with the pose, place and light you want, of yourself or a friend who agreed to it, say holding a plant pot by a shop window. Real lighting and framing are what make the result pass for a phone snapshot.

3. Swap the face in. In the Image Lab, pick GPT Image 2 (GPT Image 2.5 and Nano Banana 2 or 2.1 also work), set 9:16, attach the base photo as reference 1 and the face as reference 2, and prompt: “Same as image ref 1, but with the woman from image ref 2.” That's the playbook's one-line prompt with your character's word in the middle. GPT Image 2 at 1K medium is 10 credits; GPT Image 2.5 at the same settings is 9, plus 1 per reference.

4. Enhance it. Attach the result to Nano Banana 2.1 or Nano Banana 2 with “Enhance the image quality” (12 or 10 credits at 2K). If the pass changes too much, the playbook's longer prompt tells the model to keep the person, face, pose, outfit and background exactly as they are.

5. Run six checks. A sharp face. The usual face reference. Vertical 9:16. A phone-photo look with no glossy AI sheen. Clothes and a setting that suit your niche. Space around hands and anything else that will move. If one fails, fix the still; video only makes flaws more obvious.

Rights. Use photos you took or may edit, and a face that's fictional or belongs to a consenting adult, never a stranger's photo or someone's social media post. Spicy Studio's Terms prohibit realistic images of identifiable real people without their consent. For one face across many posts, see Consistent AI Character.

Goal: a 5-second clip of one small moment. Cost: 35 credits on Seedance 2.0 or 50 on Seedance 2.5, at 480p. Playbook page: animate workflow.

Animating is the cheapest way to get your still moving, and it works when the movement is the hook. For everyday motion like this, the playbook recommends Seedance over the Kling models.

Set it up. Open the Video Lab, choose Seedance 2.0 or Seedance 2.5 and attach your still as the First Frame. Seedance 2.5 takes the aspect ratio from the image; Seedance 2.0 starts at 16:9, so switch it to 9:16. Draft at 480p and 5 seconds. Made the still in Spicy Studio? Open it from your history and choose Animate to send it to the Video Lab with Seedance 2.5 selected.

Write the prompt like stage directions. Say what the camera does, list two to four actions in order, keep to one sequence and finish with a realism line; the playbook's animate prompt ends with “Ultra realistic. Natural movements.” For Ines:

“The camera stays completely still. She pulls a small cactus out of a cardboard box, turns it toward the lens and raises her eyebrows. Then she sets it on the counter and gives the pot a tiny pat. Ultra realistic. Natural movements.”

Sound. Generate Audio is off by default on Seedance 2.0 and on by default on 2.5, at the same price either way. Posting with a sound from the platform's library? Switch it off.

When it looks wrong. Plain image-to-video rarely looks as real as you hope, so keep the movement small. If the face drifts, cut actions. If hands warp, remake the frame with the hands fully visible. More on the models: turning images into videos with AI.

Goal: your character saying one line, word for word. Cost: 10 credits a second on Standard, at least 5 seconds. Playbook page: lip-sync workflow.

Lip-sync is the route when the words carry the clip: a tip, a confession, a short story. It's also the fallback when a clip you wanted to remake hides the face too often for motion control, because lip-sync needs only the audio.

Record the line. Use your own voice or a voiceover you hold the rights to, as an MP3 with one clear voice and little background music, trimmed to just the line. Shorter audio syncs more accurately; Spicy Lip Sync accepts up to 60 seconds. Ines's line: “If every plant you buy dies, start with a snake plant. It forgives everything.”

Make a talking frame. Repeat Lesson 2 with a base photo facing the camera, mouth closed, so the sync starts cleanly (another 20 credits).

Sync it. In Lip Sync, add the image and the MP3. Keep the playbook's default prompt, “The camera remains still at all times.”, or describe body language for a lively delivery: “She leans toward the camera, taps the pot twice on the last word and smiles. The camera remains still at all times.”

Cost. Standard (480p) is 10 credits a second and Pro (720p) 18, billed per whole second with a 5-second minimum, so a 6-second line on Standard is 60 credits.

Caption it. Plenty of people scroll with the sound off, so on-screen text matters most here. Photo and audio tips: Make a Photo Talk With AI.

Goal: your character performing a movement you recorded. Cost: 20 credits a second on Standard, at least 3 seconds. Playbook pages: motion-control workflow and, with Pro, complex motion.

Film a clean clip. Prop your phone up and record the movement yourself; for Ines, a small victory dance behind the counter. One person, the face visible throughout, a plain background, one continuous take with no cuts and a still camera. Kling 3.0 Motion Control takes driving videos of 3 to 30 seconds.

Rebuild the first frame. Screenshot the clip's opening frame and face-replace it with your character (Lesson 2), so pose and framing match.

Run it. In Character Swap, choose Kling 3.0 Motion Control, add the new frame as the character image (up to 10 MB) and the clip as the motion video. Keep Original Audio starts switched on; leave it when your voice or the room sound is part of the joke. An optional prompt steers lighting and atmosphere, and a negative prompt lists what to avoid.

Cost. Standard is 20 credits a second and Pro 27, with at least 3 seconds billed, so a 5-second remake on Standard is 100 credits. Settings in depth: the Kling 3.0 Motion Control guide. For dance clips, see the Seedance dance guide.

When the camera moves. Pans, spins, fast action and changing angles are where regular motion control struggles. Seedance can instead carry a reference video's motion, camera work and even its audio over to your character. With a reference video it bills every second of reference plus output, at 4 credits on Seedance 2.0 at 480p: a 6-second reference and a 6-second clip cost 48 credits (72 on Seedance 2.5). Small setup choices decide whether the motion and sound come through, so the tested settings and prompt are in the complex motion workflow (Spicy Pro). Model details: Seedance 2.0 guide.

Rights. Film it yourself or use footage you have permission to use. Another creator's video with your character pasted on top is still their video. If the face turns away or gets covered, switch to Lesson 4.

Goal: a short scripted clip with dialogue and sound generated in one pass. Cost: 70 credits for 10 seconds of Seedance 2.0 at 480p, or 100 on Seedance 2.5. Playbook page: sketch workflow (Spicy Pro).

With Generate Audio on, Seedance 2.0 and 2.5 produce the voices, dialogue and background sound inside the video. That opens formats that would otherwise need a second actor: street interviews, POV clips, podcast snippets, two-line comedy bits.

Keep the script tiny. A setup, a reply, a punchline, with every spoken line in quotes, in order. Seedance 2.0 runs 5–15 seconds and 2.5 up to 30, but sketches land better short, and long dialogue tends to drift. For Ines:

Interviewer: “Be honest. Do you talk to your plants?” Ines: “Every day. They just never call back.”

Build the opening shot. Make a first frame with everyone in place, here a man on the sidewalk holding a microphone toward Ines outside her shop. Attach your face reference to Nano Banana 2 (10 credits at 2K) and describe the scene in a few plain sentences, asking for an everyday phone photo.

Write the video prompt as a brief. Format and length; who's in the frame and that they stay unchanged; the action with each quoted line; the place; one simple camera behavior; the look; the sound; and what to leave out, such as music and burned-in subtitles. The Seedance Prompt Guide has a fill-in template.

Settings. 9:16, Generate Audio on, 480p for tests, and a length that fits the lines; too short cuts off the punchline.

What Pro adds. The sketch workflow has the full prompt and settings behind the playbook's street-interview example. The models are on every plan, so you can write your own sketches without it.

Goal: the clip live on Shorts, Reels or TikTok. Cost: nothing, unless you upscale. Playbook pages: the edit checklist that closes the animate, lip-sync and motion-control workflows.

Edit lightly, in any phone editor: - A caption or POV line on the opening frame, so the clip works muted. - Text or an emoji on the key moments. - A meme sound on the reaction. - A fast zoom on the punchline; it covers small glitches, too. - Skip heavy effects and filters.

Upscale only the keeper. Drafting at 480p saves credits; when a clip is worth posting, Video Upscale takes it to 1080p for 16 credits.

Label it as AI. YouTube, TikTok and Instagram all ask creators to disclose realistic AI-generated content, and Spicy Studio's Terms say every output carries machine-readable metadata identifying it as AI-generated, so don't count on a clip passing as filmed footage. Do AI Videos Get Less Views? covers each platform's rules.

Let the results pick the next clip. Try one clip per route over your first few weeks, compare how long people keep watching, and make more of what holds them. For new structures, see six viral Reels formats and AI video for social media.

Reading costs nothing; practicing costs credits, and video is where they go. Each figure is one attempt at the cheapest settings in the lessons, with no retakes.

The free 25 credits: Lesson 2's face reference, face-replace and enhance, exactly 25.

A €5 top-up (240 credits, no expiry) covers Lessons 3 to 5 once: a 5-second animation of your Lesson 2 frame (35), a talking frame and 6-second lip-sync (20 + 60), and a motion-control frame and 5-second remake (20 + 100). That's 235, with nothing left for a second try.

Spicy Basic (€10 a month, 600 credits): Lessons 1 to 5 take 260 credits, leaving 340 for retakes. Plan credits reset each billing cycle and don't roll over.

Spicy Pro (€25 a month, 1,800 credits): adds the complex motion and sketch workflows. A complex-motion clip from a 6-second reference is 68 credits with its frame (20 + 48), and a 10-second sketch 80 (10 + 70).

The honest answer to “can I make AI videos for free?”: you can learn every step and finish a first frame without paying, but motion costs credits. Plans and top-ups: pricing.

Four checks tell you whether any free course, this one included, deserves your evenings.

Current models. OpenAI announced on March 24, 2026 that it would shut Sora down; the app closed in April 2026 and the API in September 2026. A course built on Sora teaches a tool nobody can use. Look for current models such as Seedance 2.5 and 2.0, Kling 3.0 and Veo 3.1, and a recent last-updated date.

Real output. Each hands-on lesson should end with a file you made, not a quiz score.

Certificates. This course doesn't award one. If you want one for a résumé, choose a course platform that does, and check before enrolling whether the certificate costs extra.

Downloads. There's nothing to download here: the lessons are web pages that change when the models do. “Free downloads” of paid courses are frozen on the day they were recorded, and the sites offering them can carry malware.

Next steps. To grow this into a recurring character, take How to Make an AI Influencer for Free. To compare models, read Kling O3 vs Kling 3.0 vs Seedance 2.0 or browse the Spicy Academy tools track. For 18+ creators, the Free AI OFM Course covers personas and subscription platforms, and the playbook's Spicy track is coming soon.