ChatImg

What is ChatImg?

ChatImg is a multi-model AI platform for image generation and editing; users can describe new images in text form, or they can upload photos to change their style, alter their appearance, repair them, or create new versions from them.

The current account, privacy policies, and API interfaces rely on the BibiGPT service framework; however, the exact legal name of the entity behind ChatImg has not yet been made public.

Main functions

Chat-based generation and image editing

• Describe the scene, subject, composition, text, and style in an input field; once a model is selected, the image can be created.
• Upload an existing image and enter editing instructions; it can be used for style transfer, partial modifications, previewing character appearances, photo restoration, and creating visual variations.
• It allows multiple images to be uploaded at once so that the model can combine them; style reference images can also be added separately to help maintain the color scheme or visual style.
• Generation is a probabilistic process; the same prompt may yield different results, and character identities, spelling, hands, and details still require manual verification.

Multi-model switching

The workspace currently lists 12 models, covering various technical approaches from OpenAI, Google, Black Forest Labs, ByteDance, Alibaba, and xAI.

The model lineup, names, and cost per point will change in accordance with adjustments to the underlying services. Before starting mass production, it is necessary to use a small number of samples to assess stability, the accuracy of Chinese characters, consistency of the characters depicted, and the cost per image.

Style, proportion, and reference control

• The preset styles include options such as Custom, Ghibli, Lego, Figurines, Pixar, and Makoto Shinkai; more styles can also be added.
• The output ratios include preserving the original image, 1:1, 3:4, 4:3, 2:3, 3:2, 9:16, and 16:9, making them suitable for use as profile pictures, posters, images for articles, and thumbnails for short videos.
• Users can first select a scenario or style, and then add their own prompts, thereby reducing the effort required when starting with no instructions at all.
• When using brands, characters, or the styles of living artists, it is still necessary to assess on your own the risks related to trademarks, copyrights, personality rights, and platform approvals.

Scene applications and templates

• The platform offers more than 24 different options for customization, including personal color schemes, hairstyles and hair colors, makeup, profile pictures, pet portraits, logos, photo restoration, and video thumbnails.
• The scenario page precombines the input fields, prompt structure, and output layout, making it suitable for users who are not familiar with prompt syntax to try things out quickly.
• Results related to personal color, MBTI, and hand analysis are intended for entertainment or creative purposes only; they cannot replace professional advice, health assessments, or practical decisions.
• The template claims to enable a quick creation of content in the style of social media platforms, but its performance in terms of reach depends on the materials used, the quality of the content, and the characteristics of the platform; it cannot guarantee a certain level of interaction.

High-definition output, history, and external workflows

• The free version has a maximum output resolution of 1080p, while the paid plans offer 4K ultra-high-definition downloads.
• Plus offers cloud storage for image history as well as a priority processing queue, making it suitable for continuous iteration of the same project.
• Pro adds features such as converting images to videos, batch generation, and API calls, along with dedicated support.
• The page lists options for sending content directly to tools such as Notion and Obsidian; the actual formats available and the permissions associated with them should be confirmed using samples before making a payment.

Input and output

Usage tutorial

1. Open the image workspace – you can use the daily free quota right away without needing to register; logging in is required only if you want to save a history of your actions, purchase credits, or use APIs.
2. Choose text generation, image editing, or a specific scenario template, upload the necessary images, and confirm that you have the rights to work with these materials.
3. Choose the model, style, and aspect ratio, then specify the subject, environment, camera angle, colors, the elements to be retained, and the details that must not be changed.
4. Check the points required for the model before submitting; start by using a low-cost model to draft the composition, and then switch to a higher-cost model once the desired quality is achieved.
5. After creation, check the text, character features, brand elements, and details; if necessary, proceed with further modifications based on the results.
6. It is used to retain the prompt words, original materials, and records of manual edits for commercial use, and to confirm that the account package includes a commercial license.

Price packages

Subscriptions provide points on a monthly basis, while package deals are purchased once and the points in them can be used one by one without any expiration date. The payment summary on the homepage mentions \"unlimited generation,\" but the page showing the real-time prices specifies the monthly amount of points available; therefore, the payment option cannot be interpreted as allowing unlimited use of all models without any restrictions.

\"About 150 cards\" and \"About 500 cards\" are estimated using a scale of 10 points; the actual number of cards depends on the model chosen. The Alipay and WeChat Pay options indicate that renewal will not occur automatically after expiration, while the renewal procedures for other payment methods are specified on the settlement page.

One-time point package

• Generating once consumes between 5 and 50 points; complex workflows may involve multiple generations or video steps as well.
• It is not yet known whether the monthly points associated with a subscription are carried over, the validity period after cancellation, or the regional taxes and fees; these details should be confirmed before making the payment.
• ChatImg’s specific refund policy has not yet been made public; it is not possible to determine whether subscriptions and point packages can be refunded based solely on the fact that they are \"valid indefinitely\".
• Before making the payment, save the settlement page showing the package details, amount, points, and commercial benefits, so that it can be checked in case of any discrepancies.

Open API

After logging in, you can create an API Token; the credit balance for APIs and web accounts is shared, so there is no need to purchase a separate API package.

1. Prepare the accessible address of the original image or the encoded image, along with the editing prompts and model identifier, to initiate a generation task through an authentication request.
2. The interface returns the task number, status, actual points deducted, and remaining balance; the generation process is executed asynchronously.
3. The same original image is used to indicate the status of the polling task; once it is completed, the address of the generated image is retrieved.
4. Call the machine-readable price interface to obtain the cost of the current model and credits, and set budgets or a model whitelist within the script.
5. Points for failed tasks are automatically refunded, but timeout cases, duplicate submissions, and business retries still need to be recorded and verified by the caller.

• The currently available generation interface requires the original image as a mandatory input, making it more suitable for automated image editing; the parameter range for APIs that generate images from plain text has not been specified separately yet.
• The fact that APIs support batch workflows does not mean that multiple images will be returned in a single request; developers still need to arrange tasks and use polling.
• Tokens are linked to account balances and should be stored in server-side key management; they must not be included in the web interface or in public code.

Platforms and integration

Privacy and data security

The existing privacy page is based on the BibiGPT product framework; it explains that email addresses, preferences, interests, account names, and contact information may be collected, and that cookies for necessary purposes, performance optimization, marketing, and functional enhancements are used.

This policy focuses on audio and video files; it does not specify the retention period for ChatImg images, prompts, generated results, and cloud history, nor does it indicate the procedures for deleting them or whether they are used for training purposes.

• Multi-model generation implies that the images may need to be sent to the corresponding upstream processors, but the current privacy page does not list these model providers nor their respective data retention rules.
• Portrait photos may contain metadata related to the face, location, and device; irrelevant information should be removed before uploading, and the consent of the person in the photo must be obtained.
• Companies working on sensitive projects should obtain the data processing guidelines specific to ChatImg before using it; they should not rely on general privacy pages as a substitute for a compliance assessment tailored to those projects.
• The public page states that critical data is encrypted and security measures are in place, but it does not disclose information regarding independent security certifications, data storage areas, or procedures for responding to incidents.

Copyright and commercial usage notes

• The free version clearly does not include a commercial license, while the Plus, Pro versions and one-time credit packages include a commercial license.
• The commercial license included in the package does not automatically eliminate the risks related to rights associated with training materials, reference images, trademarks, portraits, and third-party characters.
• Different underlying models are still subject to their respective terms of use; it is necessary to verify the permitted uses of each model before preparing advertisements, product packaging, or delivering products to customers.
• Whether AI-generated content can be protected by copyright varies depending on the region and the degree of human involvement in its creation; important assets should include evidence of human creation and editing.
• Medical, beauty, color, and palm reading templates are intended solely for entertainment and visual preview purposes; they should not be used to make decisions related to health, finances, or any other high-risk matters.

GitHub and the open-source status

• To date, no official source code repository for ChatImg has been found, nor are there any separately released SDKs or open-source licenses.
• The open API specifications are available for public viewing, but the fact that the interface documentation is made public does not mean that the web frontend, backend, or model services are open source.
• The platform integrates multiple third-party image models, but this does not mean that the weights of these models or their training data are available to ChatImg users.

Suitable for users and typical use cases

• Social media creators: Quickly generate vertical cover images, style images, and personal profile cards.
• E-commerce and marketing professionals: Create concept sketches, logo drafts, and various visual versions of products, which are then submitted to designers for review.
• Photography and individual users: Try out different hairstyles, makeup styles, profile pictures, pet portraits, and old photo restoration.
• Designer: Compare the textual representation, stylistic consistency, and control over reference images among various models.
• Development team: Batch-edit images via asynchronous APIs that allow score sharing, and retrieve real-time costs in scripts.

Advantages and limitations

Summary

The core value of ChatImg lies in its ability to integrate various image models, reference image editing tools, scene templates, and batch processing interfaces into a single point-system framework; it is suitable for users who need to frequently compare different models or handle images automatically.

Before making a purchase, it is necessary to estimate the number of sheets required based on the actual model dimensions, and pay close attention to the scope of commercial use, image data processing, refund policies, as well as the methods for renewing subscriptions through the available payment channels.