Roma AI Docs

Create an account

Go to Register, enter your email and a password (8 characters or more), and submit. Roma sends you a verification email.

Once you're verified, log in with your email and password.

Sign in with Google

Prefer not to manage another password? Choose Continue with Google on the login or register page. Google accounts are verified automatically, so there's no email step.

What you can do without an account

The four generator pages open for anyone, no login required:

• Image
• Video
• Audio
• 3D

You can look around, browse the model pickers and tune settings. Generating needs a verified account with credits. Everything else (Workflow, Studio, History, Credits, Dashboard, Settings) needs an account too.

A brand-new account starts with 0 credits. To generate, pick a plan or a credit pack — see Plans and the Smart Custom Package.

Every generation costs credits. The price depends on the model and the settings you pick (duration, resolution, audio and so on), and it's shown before you generate.

Two separate balances

The two never borrow from each other. If you're out of video credits you can't spend image credits on a clip, and the other way round.

Where to see your balance

• Credits page: both balances, plus every batch of credits you hold.
• Dashboard: your balances, 30-day spend, generation count and most-used model.
• Settings → Billing & Credits: your plan, balances against plan size, active grants and usage history.

Batches and expiry

Credits arrive in batches, and each batch has its own expiry:

• Plan credits (Creator, Pro, Studio) last 30 days from purchase, or 365 days for the annual versions. Anything unspent when the period ends expires.
• Add-on credit packs never expire.
• Custom packages last 30 days.
• Credits granted by the Roma team, and refunds, never expire.

Refunds

• A generation fails because the provider rejects the request or errors out: your credits are refunded automatically.
• A video job never hears back from the provider: after 30 minutes Roma marks it failed, but it does not refund it automatically. The provider may have finished and billed the job anyway. If you were charged and see no result, email us and we'll review it (see Getting help).
• A generation completes successfully: it isn't refunded, even if the result isn't what you imagined.

Refunds for money you paid (as opposed to credits) are handled case by case. Email us.

Open Image. Roma has 153 image models grouped into 10 vendor tabs: FLUX · Nano Banana · Luma · OpenAI · Bria · Grok · Seedream · Recraft · Topaz · Other.

There's no separate list to memorize: the model picker is the full catalog. Choose a vendor tab, or search by name, and pick a model. Some models come in families with a Type choice (for example a Text to Image and an Edit version of the same model).

Operation types

The first choice in the composer is what you want to do:

• Text to Image: describe what you want, get an image.
• Image to Image: upload a source image and edit or transform it. Any model that requires an uploaded image lives here.
• Remove Background: a one-click shortcut to a single background-removal tool (Bria Remove Background). Upload an image, get it back without the background.

What each model asks for

Each model shows its own real settings: image size, number of images, steps, seed, output format and so on. A model that doesn't take a text prompt (some Redux and erase tools) doesn't ask for one.

GPT prompt enhancement

Flip the prompt from Classic to GPT and Roma rewrites your prompt (and negative prompt, where the model supports one) before sending it. It costs a little extra: +1, +2 or +4 credits for Low, Medium or High. You get to see the original and the enhanced version. If enhancement is unavailable, Roma falls back to your original prompt.

After it generates

Download your image, publish it to the community feed, or find it later in History. From History, Reuse sends the whole generation (prompt, settings, uploads) back to the right generator so you can iterate.

What it costs

Prices run from a few credits to a few dozen per image, depending on the model and settings. The composer always shows the current price before you generate.

Video

Open Video. Roma has 164 video models across 13 vendor tabs: Kling · Seedance · FLUX 3 · Luma · PixVerse · MiniMax · LTX · Bria · Grok · Veo · Happyhorse · Google · Other.

Every model has a type, and the picker filters by it:

• Text to Video: a prompt in, a clip out.
• Image to Video: animate a still image.
• Video to Video: start from an existing video (edit, extend, reframe, upscale, remove objects or backgrounds, motion control and more, depending on the model).

Settings depend on the model. Common ones are duration, resolution or quality tier, aspect ratio (a dropdown in the Settings menu, on the many models that support it) and audio on/off. Image-to-video models that support it add upload slots to the composer labelled Start Frame and End Frame (optional), so you can say where the clip begins and where it should land.

The price updates live as you change settings, and prices vary widely between models. The composer always shows the current price before you generate.

The same Classic / GPT prompt enhancement from Image is available (+2, +3 or +6 credits for Low, Medium or High), and for image-to-video models it can use your first-frame image as context.

Audio

Open Audio. Roma has 22 audio models from ElevenLabs, MiniMax, Google, LTX and Sonilo, split into two modes:

• Sound: music and sound effects (for example MiniMax Music, Lyria, ElevenLabs Sound Effects, Sonilo).
• Speech: text to speech (ElevenLabs, MiniMax Speech, Gemini TTS).

Speech and some music models are priced by text length or duration, so type your text first and check the estimate. Audio is paid with image credits.

3D

Open 3D. Roma has 14 3D models (Trellis, Hunyuan3D, Hyper3D Rodin, Hi3D, Meshy and others).

• Image to 3D: upload an image and get a 3D model back. Most models work this way. Meshy 6 also offers Text to 3D.
• Multi-image to 3D: Meshy 5 and Meshy V7 accept multiple images.
• Remesh, retexture and rigging + animation (Meshy): these take a 3D model file you already have (.glb, .gltf, .obj, .fbx or .stl) instead of an image. They work on the 3D page only, not inside Workflow.

3D is paid with image credits.

Workflow is a node canvas for chaining generations together. It's how you take an image into a video, or a prompt into several models, without re-uploading anything between steps.

The pieces

You add nodes from the Add Node panel in the sidebar, or by right-clicking the canvas (the same menu also opens when you drop a dragged wire onto empty canvas). You can add:

• Text Prompt and Negative Prompt
• Image Input (upload a source image)
• Image Gen, Video Gen, Audio Gen and 3D Gen: pick any model from the sidebar, organized by what a model does (Text to Image, Image to Video, Text to Speech and so on)
• Output: collects a result

Drag wires from one node's output port to another node's input port. A generation node's result can feed the next node.

Running

Your canvas autosaves about a second after each change, and undo/redo works.

Limits

• 5 saved projects per account. Delete an old one to make room.
• The Meshy remesh/retexture/rigging tools aren't available as Workflow nodes (use the 3D page).

GPT prompt enhancement

Image Gen and Video Gen nodes each have their own Classic / GPT toggle. GPT costs extra on top of the generation: +1 / +2 / +4 credits (Low / Medium / High) on image nodes and +2 / +3 / +6 on video nodes. Audio and 3D nodes don't have it.

Studio is for making a multi-shot video from a single idea. It walks you through:

1. Analyze: describe the video you want (“A 30-second luxury car commercial for Instagram, cinematic, night in Cairo…”). Roma turns it into a Brief.
2. Brief: review and edit it. You can set the platform (Instagram, TikTok, YouTube or Facebook), aspect ratio, duration (10, 15, 20, 30, 45 or 60 seconds), the number of shots (up to 50), language and audio (music, voiceover, both or none). Under the brief you'll also find a Preferred tier choice (No preference, Standard, Pro or Premier) that steers which models Roma picks for your shots, and a rough credit estimate for the whole project. It's an estimate: the real price is confirmed when you generate.
3. Elements (optional): before you build the storyboard, you can add reference elements (say, a product or a character) by uploading reference images or generating one. The panel stays visible on the storyboard.
4. Storyboard: Roma breaks the brief into shots.
5. Shots: Roma routes each shot to a model based on your preferred tier. Before any credits are spent, a Confirm generation dialog lists each shot, the model chosen and its price, and you approve the total. You can generate one shot at a time or several at once.
6. Assemble: stitch the shots into a final video. Voiceover lines and sound-effect cues have their own editors. Assembling re-packages shots you've already paid for, so it's free, apart from background music when your audio setting includes it — in that case the estimated music cost is shown next to the button.

Your projects live on the Studio page, and you can reopen any of them.

The three plans

Every plan includes access to all 350+ AI models, Workflow, and the affiliate program. Annual plans work out about 17% cheaper than paying for twelve months one at a time. Studio adds nothing on price-per-credit: credits cost the same across all three, so pick the plan by how much you'll use.

Add-on credit packs

One-time top-ups that never expire:

The Smart Custom Package

Need more than Studio? Build a Custom Package on the Pricing page:

1. Pick what you make. Choose video types (Social & UGC clips, Product & ad videos, Cinematic & premium) and image types (Quick social images, Marketing & product photos, Personal & creative use). Each shows roughly how many credits one video or image costs.
2. Set your monthly volume with the two sliders (up to 400 videos and 3,000 images).
3. See the total in credits and price (USD, with an EGP equivalent).
4. Continue to payment.

A custom package starts at 14,999 credits, which is Studio's own level. For less than that, the regular plans are the better fit. The largest package you can buy in one go is capped at 65,000 EGP (the most a single transfer can carry). For more, email us. Like the plans, a custom package is a one-off purchase whose credits last 30 days.

How to pay

Roma takes payment by local transfer, verified by our team. There are no card payments.

1. Choose a plan, pack or custom package. You'll land on the payment page.
2. Pick a method: InstaPay or Wallet.
3. Send the amount shown (in EGP) using the link or number on the page. Your order is held for 60 minutes.
4. Upload a screenshot of your receipt.
5. You'll see one of three results: Approved (credits added right away), Under review (we're checking it and you'll be notified), or Could not verify (contact us if you think that's a mistake).

Use the EGP toggle on the Pricing page to see prices in Egyptian pounds. The amount you actually pay is calculated when you place your order.

Every Roma account can earn by referring people who become paying customers.

1. Get your link. Open Affiliate (it's in your profile menu) and choose Generate Referral Code. Your link is the Roma address with ?ref= and your code on the end. Copy it from the same page and share it anywhere.
2. Attribution. Someone who signs up through your link within 30 days is counted as your referral.
3. Commission. You earn $1.00 for each qualified referral at Level 1, and the rate rises as you grow. Moving up needs both the referral count and the revenue total:

Payouts. The minimum payout is $100. Your Affiliate page shows your pending balance and how far you are from the minimum. When you reach it, request a payout by Visa, InstaPay or Wallet. Our team reviews each request. Any transfer fee is deducted, and your payout history shows the net amount.

Account settings are under Settings.

Forgot your password?

On the login page, choose the forgot password link and enter your email. If it's registered, Roma emails you a single-use reset link. Open it and set a new password (8 characters or more).

Changing your password

Settings → Security → Change Password. Enter your current password and the new one.

Sessions

Roma allows one active session per account. Settings → Security shows your current session and recent history, and Log out other sessions signs everything else out.

Two-factor authentication isn't available yet.

Deactivating vs deleting

Deactivate (Settings → Danger Zone) is temporary. Your profile and posts stop being visible to others, but nothing is deleted. Log back in any time to reactivate.

Delete is permanent and can't be undone:

• You'll be asked for your current password and to type DELETE.
• Your posts are unpublished right away, and your username, display name, avatar and bio are removed.
• Comments and likes you left on other people's content stay, shown as “Deleted User”, so their threads don't break.
• Billing records are never deleted.

Check the Support page first — it has answers to the questions we hear most, grouped by topic. If it doesn't cover what you need, email romaaieditor@gmail.com directly (the same page has a quick contact form that opens your email app pre-filled).

To help us help you, include:

• the email on your account
• what you were doing and which model or tool you used
• for a charge with no result, roughly when you generated it
• for a payment, your order details and receipt