DFocus's First Character Project | D-mei × Steve F
How We Built Our AI Brand Characters
From on-premises AI image workflows to actual applications on the website, D-mei and Steve F the pug have become DFocus's AI ambassadors and brand mascots. This time, we integrated character design, clothing adjustments, facial compositing, dual-character pairing, and website deployment into a single workflow, allowing the characters to truly participate in introducing the company's services.
release:| Production and case study provided by: DFocus (鼎豐整合行銷有限公司)
Let AI play a role in building a sustainable brand face for businesses.
Creating a brand persona doesn't necessarily require investing in real-life spokespeople, studios, and repeated photoshoots up front. DFocus used its own website as an experimental ground, first establishing its own AI characters, "D-mei" and "Steve F," and then using transparent images of them in different poses in service introductions, case studies, and collaboration project discussion sections. The characters are part of the content guidance and brand visual identity; the company still relies on a real-life team for service and delivery.
The "first-ever character project" here refers to DFocus's proprietary dual-character project. This report documents our production methods and deployed footage, serving as a practical reference for companies evaluating the creation of AI spokespeople.
What roles do the AI spokesperson and the mascot play?
D-mei: Making our services more approachable
The brand establishes its identity through a fixed face, hairstyle, crown, gray suit, and red tie, then arranges gestures such as pointing, introducing, displaying, and waving according to the content. Different pages use the same character in different contexts, ensuring a consistent brand image for the company's services.
Steve F: A brand mascot that coordinates well with the spokesperson
The pug, dressed in a black suit, purple tie, and brand-identified accessories, echoes the service theme by holding charts, scanners, or making introductory gestures. Characters and mascots can appear separately or as a pair, extending to case studies and future video projects.
Less budget for live-action filming, more character footage that can be repeatedly developed.
Having your own AI spokesperson can reduce reliance on external endorsement offers, shooting locations, and scheduling. Once the character is set up, businesses can extend the poses and visuals according to the needs of new services, events, or products, reducing the work of creating something from scratch each time.
Actual implementation still requires listing costs for character design, hardware or model services, image generation and retouching, character consistency checks, website integration, and subsequent updates. DFocus will first confirm the intended use and scope of delivery of the materials; this case study does not use real-person endorsement pricing for cost comparison, nor does it equate character visibility with inquiries or sales results.
AI Spokespersons, AI Mascots, AI Models: Giving Corporate Visuals a Clear Purpose
DFocus can assist in planning AI visual workflows for proprietary brand characters, personas and clothing design, mascot pairings, and product contextual images. These applications use images and videos to express the brand; when documents need to be accessed, business opportunity data needs to be created, or tool operations need to be performed, a separate AI Agent workflow is planned .
AI Spokesperson | A Fixed Face for Brand Communication
With consistent facial features, clothing, and identification characteristics, the website's service guidance, event visuals, and audiovisual planning were extended. D-mei was the character actually deployed on the website for this purpose.
AI Mascot | A Brand Partner That's Easier to Remember
Using Steve F, a pug, as a character, the analysis, inspection, and introduction gestures are applied to service descriptions and case study pages, based on the content.
AI Models | Scenario Experiments with People, Clothing, and Products
Starting with a fixed character, we edited their clothing, poses, and product pairings. The example screenshots already show the adjustments to the character and clothing; the final product images need to be checked separately for materials, proportions, details, and usage requirements.
AI Agent | Integrating Tools and Resources into Tasks
For tasks such as document retrieval, prospect list research, or website alerts, design data sources, execution permissions, recording and manual review, and plan them separately from image roles.
What a character project can deliver
The D小妹 and Steve F case shows a practical path from reference images and character edits to brand assets used on a live website. For a client project, agree on the following deliverables before production:
- Character reference: approved face, hair, clothing, proportions and mascot details.
- Asset set: agreed poses, expressions, transparent cutouts and desktop/mobile sizes.
- Human review: identity consistency, hands, edges, brand marks, image resolution and intended use.
- Placement: approved website sections, readable text areas and responsive image behavior.
- Usage and handover: source-material permissions, tool/model terms, final files and revision scope.
Static character assets and video production are separate deliverables. A still image or video-generation workflow screenshot is not a finished video. Additional clips should be reviewed and approved before publication.
Not just remaining in the output folder: D-mei and Steve F have arrived on the website.
DFocus places transparent character images alongside text, along with color blocks and collaborative sections, and adjusts the size, cropping, and reading flow for both desktop and mobile devices. The mobile version also offers compressed WebP format to balance character recognition and download volume.
The above is an application used on DFocus's own website, and is not the work of the listed clients who hired these characters to represent them. The website presentation is subject to continuous optimization and adjustment.
From character creation to brand launch: The actual production workflow of AI spokespeople
DFocus used its local ComfyUI and Qwen Image 2.1 workflow to process character references, adjust clothing, composite faces, incorporate mascots, and coordinate with brand visuals before integrating the finalized character into the short-form video workflow. The screenshots below document the node settings and output previews at each stage, serving as practical verification of the on-premises AI image application. They demonstrate the production process and stage results, but do not represent client outcomes or guarantee complete consistency with the generated results.
- Character creationPrepare character reference materials and set the character's appearance and generation direction.
- Clothing AdjustmentAdjust clothing and visuals according to the brand context, while checking character identification elements.
- Face compositingIntegrate reference faces with the new look, and review consistency, expression, and naturalness step by step.
- Mascot IntegrationAdd the mascot to the character group and adjust its proportions, clothing, and composition.
- Complete Brand Character SystemIntegrate characters, mascots, products, and brand visuals into extensible AI spokesperson materials.
- Video WorkflowIntegrate image characters into the video generation process; currently showcasing the workflow and stage screenshots, video examples will be added after the final product is completed.
The current page displays the workflow of character creation, clothing adjustment, face compositing, mascot integration, and short-form video recording; the complete video will be added later. During the implementation process, the character portrait, reference photos, product materials, and model/platform licensing scope will be confirmed first, and the generated results will be checked by a person before delivery.
Common Questions about Creating AI Spokespersons for Businesses
Where can AI spokespeople be used in businesses?
It can be used for website service introductions, project inquiry sections, brand mascots, and product contextual visuals and audio-visual planning. DFocus has deployed D-mei and Steve F on its own website service page, and configured different presentation methods for mobile and desktop computers.
What verifiable information is available regarding DFocus's AI spokesperson cases?
This page provides workflow screenshots, character creation stages, output previews, and a link to the actual website application for ComfyUI and Qwen Image 2.1. Image deployment is complete; the short-form video currently provides workflow and stage previews, which will be added after the official video is completed.
Can AI spokespeople save on the costs of human spokespeople and filming?
If the purpose is to build a proprietary brand persona, the AI workflow can reduce the need for in-person invitations, studio sessions, and repetitive shooting. However, design, generation, retouching, hardware or service fees, review, and website integration costs still need to be considered. Actual costs will be confirmed in your quote based on requirements and the scope of delivery.
How can we make an AI spokesperson look like the same person every time?
First, establish character references, fix clothing and identifying features, and then create poses, expressions and scenes in stages. Through reference images, node settings and manual comparison to check faces, body shapes, hands and accessories, the qualified results are organized into extensible material specifications.
Are AI spokespeople and AI agents the same thing?
AI spokespeople are applications of brand personas and visual media; AI agents focus on tool integration and task workflows. This case study is an implementation of the image workflow within DFocus AI implementation services; document processing, business development, and website monitoring can be planned separately.
What do businesses need to prepare before implementing this?
Prepare brand positioning, character style, platforms to be used, authorized reference materials, delivery dimensions, and budget; if referencing real people, the scope of use must also be confirmed. A set of character materials for the website can be created first, and then expanded to product scenarios or videos based on the review results.
What is the difference between an AI spokesperson and an AI agent?
AI spokespeople are brand characters used for external communication and can be used on websites, in images, or in videos; AI agents are workflow systems that connect data and tools and perform tasks according to permissions. DFocus's D-mei and Steve F are brand visual applications; AI sales hunters, document retrieval, and alarm summaries are AI agent applications, which can cooperate but have different functions.
How to choose AI mascots, AI models, and AI spokespeople?
For a stable brand image, an AI spokesperson can be planned; for conveying brand personality through animals or fictional characters, an AI mascot can be planned; for presenting visuals of clothing, merchandise, or scenarios, AI models can be evaluated. DFocus has demonstrated workflows for character creation, clothing adjustment, facial synthesis, and mascot pairing. Actual delivery will depend on the materials, intended use, and trial results.
Can DFocus help create AI-generated images of models and products?
We can discuss visual prototypes of AI-generated characters, clothing adjustments, and product scenarios. This case study provides screenshots of the workflow for character and clothing embellishment and product pairing; before official use, we will still check character consistency, hands, product appearance, brand recognition, and delivery dimensions. Companies can first select a set of authorized materials to confirm the scope of the prototype and acceptance criteria.
Will the AI spokesperson images themselves answer customer questions or perform tasks automatically?
Static character images will not automatically answer questions or perform tasks. If interactive dialogue, order inquiries, or operating system functions are required, a separate dialogue interface, data sources, tool permissions, and manual confirmation processes must be established. Character appearance can be part of the interface, but it cannot replace the system design of the AI Agent.
Want your company to have an AI spokesperson?
Let's start with a practical application with clear acceptance criteria: website service guidance, product scenario illustrations, brand mascot, or short-form video project. Please provide brand information, authorized materials, desired scenarios, and available budget. DFocus will discuss suitable workflows, pilot scope, and delivery methods.