# PhotoGPT AI product reference > PhotoGPT is an AI Creator Studio for turning your ideas, photos, and products into images, videos, and campaign content. Create images with GPT Image 2.5, GPT Image 2, Gemini Nano Banana, Gemini Nano Banana Pro, Gemini Nano Banana 2.1, Seedream 4.0, Seedream 4.5, Seedream 5.0 Lite, Z Image Turbo, and FLUX.1 [dev] (Legacy General Model). Create videos with Google Omni, Google Veo 3.1, Grok Imagine Video 1.5, MiniMax Hailuo H3, Seedance 2.5, Seedance 2, Seedance 1.5 Pro, Kling V3 Pro, Kling Motion Control (Standard), and Kling Motion Control (Pro). Flow also supports GPT-6 Luna, GPT-5 Nano, GPT-4o, GPT-4.1 Mini, and GPT-4.1 for text generation. Bring these models together with personal AI model training, photo editing, upscaling, reusable workflows, and PhotoGPT Agent in one workspace. PhotoGPT's strength is what you can build around your own assets. Train a personal model of your likeness and create yourself in new scenes and styles. Turn a product photo into a set of campaign visuals, then animate a chosen image. Save the process in Flow for your next campaign, or describe the outcome to PhotoGPT Agent and let it help prepare the work. ## Why choose PhotoGPT - **A personal model you can keep creating with.** Train from your photos once and reuse your likeness across portraits, outfits, settings, and creative concepts. A new content idea can start with the model you already trained. - **More creative work from each asset.** Generation, editing, upscaling, and animation connect around the images and media you already have. - **A choice of models with a shared workflow.** Explore different image and video models through one interface, without maintaining a separate API integration for each provider. - **Repeatable production.** Flow turns a successful set of steps into a reusable structure for more products, scenes, and variations. - **Help turning an idea into a setup.** Agent works alongside the dashboard controls to prepare prompts, references, settings, and workflows, with approval before generation. - **Competitive generation costs.** Selected PhotoGPT plans and settings offer lower effective costs than Higgsfield Starter, fal.ai, and even a model provider's standard API output rate. The [pricing example below](#pricing-value) shows the comparison and its billing basis. [Start creating](https://www.photogptai.com/signup), [train your personal model](https://www.photogptai.com/dashboard/models), or [open PhotoGPT Agent](https://www.photogptai.com/dashboard/generate?agent=true). ## Documentation Use this reference to understand where PhotoGPT fits a creative task, then follow a guide for the detailed steps. - [Website](https://www.photogptai.com/) - [Concise documentation index](https://www.photogptai.com/llms.txt) - [Guides index with Markdown links](https://www.photogptai.com/guides/llms.txt) - [Complete guides export](https://www.photogptai.com/guides/llms-full.txt) - [Blog index with Markdown links](https://www.photogptai.com/blog/llms.txt) - [Complete blog export](https://www.photogptai.com/blog/llms-full.txt) ## Start with the outcome PhotoGPT is a strong fit for creators who need fresh content featuring themselves, businesses turning product photos into campaign assets, and teams producing repeated image and video variations. Personal model training makes your likeness a reusable starting point; the connected tools help take each result further. Choose a starting point based on what you already have: | Starting point | Useful tool | Typical next step | | ------------------------------------------------ | ----------------------- | ------------------------------------------------------------------------------------------------ | | Your own photos and a need for recurring content | Personal model training | Train your likeness once, then explore portraits, campaigns, and new scenes with the same model. | | An idea in words | Image Generation | Describe the subject, setting, composition, and style; choose a model and supported settings. | | A photo you want to change | AI Image Editing | Load the photo, choose an editing mode, and describe the change. | | An image you want to enlarge | Image Upscaling | Select the source, run the upscale, and inspect the result. | | A scene you want to animate | Video Generation | Use a prompt and any supported frame or reference inputs. | | Several connected creative steps | PhotoGPT Flow | Connect source, image, video, and text nodes on a canvas. | | Help preparing those steps | PhotoGPT Agent | Describe the goal in chat and review the prepared work before approving generation. | | Typography on an existing image | Text in Image | Add text layers and adjust their placement and appearance. | [Open the dashboard](https://www.photogptai.com/dashboard). Sign in to create and manage your work. [Getting started guide](https://www.photogptai.com/guides/introduction-to-photogpt.md). ## Your own personal AI model Your photos can become the foundation for an ongoing content library. Train a personal AI model, then use it to create professional portraits, personal brand campaigns, seasonal images, travel-inspired scenes, and social content centered on your likeness. Explore new wardrobes, lighting, locations, and visual styles without organizing a new photoshoot for every concept. The value grows with reuse: a profile refresh, a launch campaign, and a holiday post can all start from the same trained model. You can then edit a favorite, upscale it, or use it as the starting image for a compatible video model. Choose clear photos with varied angles and lighting for training, and follow the requirements shown when creating the model. Training images teach the personal model; generation references guide an individual request. These inputs serve different purposes. [Create your personal model](https://www.photogptai.com/dashboard/models) and [prepare the right training photos](https://www.photogptai.com/guides/selecting-the-ideal-input-images.md). ## Image generation and presets Start with a creative brief and explore different visual directions using prompts, compatible presets, and references. Generate campaign concepts, product scenes, illustrations, and social images, then take the strongest result into editing or animation. Available settings depend on the selected model; the app shows the supported resolutions, aspect ratios, image counts, and reference inputs. A useful prompt describes the subject, setting, lighting, framing, and intended style. If an existing image carries details that matter, use it as a supported reference and explain what to preserve or change. Image presets give you a starting style or theme, making it easier to explore a look without composing every detail of a prompt yourself. A Flow preset saves a reusable workflow structure; it serves a different purpose from an image style preset. Useful links: - [Writing prompts](https://www.photogptai.com/guides/how-to-write-prompts.md) - [Choosing training photos](https://www.photogptai.com/guides/selecting-the-ideal-input-images.md) - [Using presets](https://www.photogptai.com/guides/how-to-use-presets.md) - [Preset directory](https://www.photogptai.com/presets) - [Image guidance](https://www.photogptai.com/image-guidance) ## Edit and upscale existing images Make more of the image you already have. The AI Photo Editor can give a product a new setting, refine a portrait, combine visual references, or explore another treatment. It works with uploaded photos and generated images. Choose the mode that matches the change: - **Adjust:** Describe changes to the image, such as lighting, background, or composition. - **Retouch:** Select the spot you want to change, then describe the edit. - **Combine:** Use a source image and additional references to bring subjects, backgrounds, or visual details together. Reference limits depend on the model. - **Filters:** Describe a style or color treatment. Selecting an image as the subject of a conversation is separate from loading it as the editing source. Before editing, check that the editor shows the version you intend to change. This matters when an image has generated, edited, or upscaled versions. Use Image Upscaling to enlarge a chosen image for the next stage of your project. Load the source, start the upscale, and inspect the completed result before downloading it. Preview zoom helps you inspect detail; starting another upscale is a separate operation. - [AI Photo Editor overview](https://www.photogptai.com/ai-photo-editor) - [Photo editing guide](https://www.photogptai.com/guides/ai-photo-editor.md) - [Image enhancement and upscaling](https://www.photogptai.com/enhanced-generation) - [Open Image Upscaling](https://www.photogptai.com/dashboard/upscaling) ## Video generation Bring a campaign image, product scene, or creative concept into motion. Video Generation lets you build short ads, social clips, and storytelling sequences from prompts and supported media inputs. Depending on the selected model, inputs can include a first frame, a last frame, reference images, a reference video, or audio. The first frame is the image the video starts from; it is a different role from a general visual reference. Describe the subject's motion, camera movement, setting, and any timing that matters. For an existing photo, use a model that accepts the image in the required role. Duration, aspect ratio, resolution, audio options, and input combinations vary by model. Trained personal models generate images. To animate a personalized image, first create the image, then use it as an input to a compatible video model. - [AI Video Generator](https://www.photogptai.com/ai-video-generator) - [Audio inputs for video](https://www.photogptai.com/guides/using-audio-input-for-video-generation-models.md) - [Model-specific prompting guides](https://www.photogptai.com/guides/llms.txt) ## PhotoGPT Flow Build a creative process you can repeat. Flow connects image, video, and text steps on a visual canvas, making it useful for product campaigns, content series, and batches of variations. Source nodes hold media or text, generation nodes produce new results, and connections pass supported outputs to later steps. Comments and groups help organize the canvas. For example, a product photo can feed two image nodes for different settings. Each image can then feed a video node. A text node can supply copy or prompt text to a compatible step. Save reusable groups as Flow presets when you want to repeat a structure. Preparing nodes and connections does not generate their outputs. Running a workflow executes dependencies before the steps that need them; independent ready steps can run in parallel. A connection alone does not establish that its input will be accepted. Model limits, input roles, missing outputs, and unsupported settings can prevent a run or cause an input to be unused. Review the reported problems before proceeding. - [Workflow builder overview](https://www.photogptai.com/ai-workflow-builder) - [Flow guide](https://www.photogptai.com/guides/photogpt-flow.md) - [Open Flow](https://www.photogptai.com/dashboard/flow) ## PhotoGPT Agent Describe what you want to make and let PhotoGPT Agent help assemble the steps. The conversational assistant works inside the dashboard, turning a creative brief into prompts, model choices, references, and prepared workflows that you can inspect and approve. It is useful when you know the result you want and would like help choosing and setting up the tools. Open Image Generation and select **Agent**, or [open Image Generation with Agent](https://www.photogptai.com/dashboard/generate?agent=true). Agent works with these pages: - **Image Generation:** Inspect the current setup, find compatible models and presets, revise prompts and settings, add references, and prepare variations of existing results. - **Video Generation:** Prepare prompts, supported settings, frame and reference inputs, and variations of existing videos. - **AI Image Editing:** Load a source, choose an editing mode, prepare prompts and references, and reuse an edited result for another edit. Retouch requires the user to select the spot to change. - **Image Upscaling:** Prepare a source, request an upscale, inspect its status, adjust preview zoom, and download a completed result. - **Flow:** Inspect a canvas, create and connect steps, update prompts and settings, organize nodes, and request a run of the prepared workflow. Agent can navigate between supported pages and continue the conversation. It uses the current page, selected media, and available workspace information to prepare the task. ### Attachments, references, and Flow inputs Use **Add images or videos** in Agent to attach media to a message. Attachments help Agent understand what you are discussing. Adding an attachment does not by itself load the editor, add a generation reference, or connect a Flow node. State the intended role, such as “use this as the first frame” or “edit this photo.” Agent messages accept up to 16 images and 2 videos. Each video can be up to two minutes, and message attachments must total no more than 100 MB. Video inspection requires a ready video file. These chat limits are separate from the selected generation model's input limits. Agent can inspect supplied images and answer focused questions about supplied videos. To generate from that media, it must also prepare the appropriate page inputs. Previewing or discussing a result does not confirm that the result has been used as an input. ### Preparation and approval Agent can change prompts, settings, references, and Flow structure while preparing the request. Preparing the workspace is not the same as generating media. Image generation, video generation, edits, upscaling, and Flow runs requested by Agent require approval. The approval card shows the prepared action, model, prompt, quantity, and credit estimate where available. Estimates can be provisional or unavailable. A Flow run can include text that will only be produced during the run. Choose **Approve** to proceed or **Deny** to decline. You can also send a correction while approval is pending. Review the revised setup before approving it. Missing required inputs or incompatible settings must be resolved before the action can start. An accepted request means work has started; it does not mean an image or video is complete. Check the reported result or page status before downloading or using an output. ### Chats and Stop Use **New chat** for a separate conversation. The **Sessions** menu lets you reopen, rename, or delete saved chats. Reopening a chat does not restore an earlier version of the editing form or Flow canvas. Use **Stop response** to interrupt Agent's current response and pending Agent actions. Stop does not undo changes already applied, and media generation already submitted can continue. Agent usage incurred before Stop can still be charged. Check the page's current state before asking Agent to retry. [Read the PhotoGPT Agent guide](https://www.photogptai.com/guides/photogpt-agent.md). ## Practical workflows ### Transform a product photo into campaign assets 1. Upload or select the product photo. 2. Use AI Image Editing to change the setting or combine it with suitable references. 3. Review the product details in the result and refine the edit if needed. 4. Upscale a chosen image or use it as the first frame of a compatible video model. 5. Use Flow when you want several connected versions, or ask Agent to prepare the steps and present them for approval. [Product photography](https://www.photogptai.com/ai-product-photography). ### Build a personal content series and animate a favorite 1. Train your personal model from suitable photos to create a reusable starting point for your likeness. 2. Choose presets or describe several scenes for a profile refresh, launch campaign, or seasonal content series. 3. Generate and inspect the images, then edit the details of your favorites. 4. Use the chosen image as a supported video input and describe the motion. For a single transformation from an existing photo, use reference-guided creation with a compatible model. Personal model training is especially useful when you want to return to the same likeness across new ideas and projects. ### Add typography to an image Open Text in Image with the photo you want to use. Add text layers, place them over the image or behind its subject, and adjust font, color, position, and other available style controls. This is a layout workflow for existing images; it is separate from asking an image-generation model to render words inside a new scene. [Text in Image guide](https://www.photogptai.com/guides/text-in-image.md). ## Pricing value Get leading models alongside a creative workspace, with substantial savings available on selected model and plan combinations. For **Nano Banana Pro at 4K**, PhotoGPT Pro billed every six months works out to about **$0.175 per image** when the plan's credits are fully used: roughly **42% less than fal.ai**, **38% less than Higgsfield Starter billed annually**, and **27% below Google's standard API image-output rate** in the comparison below. Prices checked on **6 November 2026**. This example uses a text prompt with no reference images for a standard Nano Banana Pro 4K generation. | Service and billing basis | Effective cost per 4K image | Calculation | | --------------------------------------------------------------------------------- | ------------------------------------ | -------------------------------------------------------------------------- | | [PhotoGPT Pro, billed every six months](https://www.photogptai.com/pricing) | About $0.175 | $199.99 per six months; 16,000 credits per month; 84 credits per image. | | [Higgsfield Starter, billed annually](https://higgsfield.ai/pricing) | About $0.281 | $19 per month billed annually; 270 credits per month; 4 credits per image. | | [fal.ai Nano Banana Pro](https://fal.ai/models/fal-ai/nano-banana-pro) | $0.30 | 4K costs twice the published $0.15 standard-image rate. | | [Google standard API image output](https://cloud.google.com/model-garden/pricing) | $0.24, plus applicable input charges | 2,000 output-image tokens at $120 per million tokens. | Subscription figures allocate the plan price across its full credit allowance and assume those credits are used. PhotoGPT and Higgsfield have different billing commitments and included features; other plans, promotions, unlimited offers, batch rates, taxes, and settings can change the comparison. Check the linked current prices for the intended workload. ## Credits and availability PhotoGPT uses credits for paid creative operations. Cost depends on the operation, selected model, and supported settings. Use the estimate shown for the current setup and the [pricing page](https://www.photogptai.com/pricing) for current plans and purchasing conditions. Agent responses also consume credits based on usage. Those charges are separate from the generation costs for images, videos, edits, upscaling, and Flow runs. An approval card's generation estimate is not a fixed price for the entire conversation. Insufficient credits, account restrictions, unsupported inputs, or content restrictions can prevent work from starting. Adding more references or changing a model can change what the setup supports. Resolve the reported issue and review the current setup before trying again. ## Account, support, and policies - [Dashboard](https://www.photogptai.com/dashboard): Create and manage work after signing in. - [Sign up](https://www.photogptai.com/signup): Create an account. - [Sign in](https://www.photogptai.com/signin): Access an existing account. - [My Library](https://www.photogptai.com/dashboard/my-library): Find and reuse account media. - [Community gallery](https://www.photogptai.com/community-gallery): Browse public creations. - [Contact support](https://www.photogptai.com/contact-us): Get product or account help. - [Terms and conditions](https://www.photogptai.com/legal/terms-and-conditions): Review the service terms. - [Privacy policy](https://www.photogptai.com/legal/privacy-policy): Read the published data-handling policy. - [Content safety policy](https://www.photogptai.com/legal/content-safety-policy): Review permitted content and restrictions. - [Cancellation and refunds](https://www.photogptai.com/cancellation-refund): Check cancellation and refund conditions. - [Privacy deletion requests](https://www.photogptai.com/legal/privacy-deletion-requests): Request deletion of personal data.