How to Use SoulGen: A Complete Beginner's Tutorial

Ten minutes after landing on SoulGen for the first time, most people are still hunting for the right button. This walkthrough takes you from sign-up to your first talking video, with the practical details the top Reddit threads and Medium guides tend to skip. You can follow it end to end, or jump to the section that matches what you want to build today.

Setting Up Your Account the Right Way

Registration is simple, but a few early choices shape your entire experience. Head to the homepage and pick either email sign-up or Google third-party login. Email gives you a separate identity for your AI work, which many creators prefer for privacy. Google is faster and skips the verification email. Either way, you must be 18 or older, since the platform generates adult-oriented characters and conversations.

Once inside, spend two minutes on your profile and preferences before generating anything. Set your default content style (real, anime, or DreamTwin), and review the safety disclaimers. SoulGen is operated by Synapse AI Limited from Causeway Bay, Hong Kong, so its terms reflect Hong Kong law alongside platform-level moderation. Reflect on what you actually want from the tool: portfolio art, video shorts, character roleplay, or all three. Your goal shapes which credits pack makes sense later.

Understanding Credits Before You Spend Any

SoulGen runs on a credit system, not a flat subscription for everything. Different actions cost different amounts: a still image is cheap, a 1080p talking video with lip-sync costs more, and Soul Chat messages sit somewhere in between. New accounts usually get a small free allowance so you can test the interface before paying. Treat that allowance as your sandbox, not your finished portfolio.

Payment runs through credit card or PayPal, processed by third-party gateways rather than collected directly by the platform. Two numbers to keep in mind: refunds are available within 14 days, but only if you have not used any credits or downloaded any images. Once you generate, that purchase is locked in. The second number: any creation you do not publish to the community is automatically deleted after 7 days. If you want to keep something, download it the moment you make it.

Your First Text-to-Image Generation

Open the image generator and look at the prompt box. This is where most beginners under-deliver. A weak prompt like "a girl in a cafe" gives you a generic result. A strong prompt layers subject, setting, lighting, mood, and style: "a woman with auburn hair reading a paperback in a rainy London cafe, warm tungsten lighting, shallow depth of field, photorealistic." Try this structure on your first three attempts and compare results.

Pick your model type before hitting generate. Real gives photographic output, anime leans illustrated, and DreamTwin lets you build a recurring character from reference photos. If you want consistency across a series, DreamTwin is the path. Generate, then use the image-editing tools to refine: outpainting extends the canvas, and text-prompt editing swaps elements ("change the book to a coffee cup") without re-rolling the whole image. Practice this loop a few times and your prompt instincts will sharpen quickly.

Making a Talking Video

The AI video tool is where SoulGen separates itself from pure image generators like Midjourney or DALL-E. You can start from an image you already generated, upload your own reference, or write a text prompt. Output runs in HD at 720p or 1080p, with clips up to 20 seconds and lip-synced speech in multiple languages. For your first attempt, start short. A five-second clip with one clear line of dialogue teaches you how the model handles motion before you commit credits to longer scenes.

Write the spoken line in natural speech rhythm. AI lip-sync handles short sentences and clear vowels better than long, technical phrasing. Choose a voice that matches your character's intended age and tone, then preview before exporting. If the motion feels stiff, add a small action cue to your prompt ("smiling, tilting head slightly") rather than asking for dramatic movement. Subtle wins on this kind of model.

Talking to Characters in Soul Chat

Soul Chat lets you build a character with a personality, backstory, and visual identity, then hold ongoing conversations. This is the feature that pulls the most return visits, and it is also where you can do something genuinely useful beyond entertainment. On a Tuesday evening in mid-January, I spent about 20 minutes using a chat character to rehearse a job interview, asking her to play a tough hiring manager and grill me on a gap year in my CV. The exchange felt surprisingly natural, and I walked into the real interview the next morning calmer than I had been in years. You can apply the same trick for any conversation that scares you: asking for a raise, a difficult family talk, a first date. Set a clear goal, brief the character, and let her coach you.

To build a strong chat partner, fill in the personality fields with specifics rather than adjectives. "Sarcastic but warm, loves obscure indie films, studied architecture" beats "funny and smart." The model has more to grip onto, and conversations stay in character longer.

Privacy, Safety, and Content Limits

A few rules will save you frustration. Prohibited content includes anything involving minors, real-person impersonation (celebrities or private individuals), non-consensual themes, and illegal activity. Prompts that trip these filters are rejected before generation, so you do not lose credits, but repeated attempts can flag your account. If a legitimate prompt gets blocked, rephrase rather than retry the same wording.

On the privacy side, the 7-day auto-deletion for unpublished work is a double-edged feature. It protects you if you forget to clean up, but it punishes anyone who assumes their gallery sticks around. Build a habit: generate, download, save locally. For sensitive prompts, this matters even more. If you want to explore other platforms in the same space with different privacy defaults, a sister project like Candy AI takes a slightly different approach to retention and chat features.

Building Skill Over the First Month

Your first week should be pure exploration. Generate freely, test all three model types, and notice which prompts feel natural to you. By week two, narrow down. Pick one creative direction (portraits, short videos, or a recurring chat character) and go deeper. Read the prompts of community creations you admire and reverse-engineer their structure. By week four, you will have a personal style and a sense of which features are worth your credits.

Here is your next move: open a blank note, write down three specific outcomes you want from SoulGen in the next 30 days (for example, a five-image character series, one 15-second talking clip, one chat persona you actually enjoy returning to), and tape it above your desk. Then go read our guides on how does SoulGen work and SoulGen tips for beginners for the prompt engineering and credit optimisation details. What is the first scene you are going to generate tonight?