Personal project · In private use · Private source

ThaiGlow

A private AI studio that turns product photos into styled images, short videos, and Thai selling copy, with a simple credit budget.

ThaiGlow - selecting the style of the pictures to generate
ThaiGlow - selecting the style of the pictures to generate

About ThaiGlow

I built ThaiGlow for my girlfriend, who needed product images and short videos for her social-selling posts. Inspired by Aidentika, the idea was to give her a simple tool she could use from her phone: upload a product photo, choose a style, and create material for her next post.

The app brings product-image generation, short vertical videos, and Thai sales-copy drafts into one workflow. It offers 16 scene styles, several quality options, and multiple image variations. Optional product details help shape the generated titles, selling points, and post text for TikTok Shop, Shopee, Facebook, or Instagram. The text remains editable before she copies or shares it.

Image and video generation run in background queues, so she can close the app after submitting a job and return to the results later. Completed media is saved in a shared gallery, while history keeps earlier work available to revisit. Watermarks, product infographics, and mobile sharing add the finishing steps around generation.

The interface supports English and Thai, light and dark themes, and installation as a web app. Everyday controls focus on styles, quality, and results. A separate administrator view lets me adjust the supported models and prompts without putting those settings into her normal workflow.

The credit system started as a joke between us and became a practical allowance for experimenting with paid generation. Each image or video uses a fixed number of credits; unsuccessful jobs return credits, and I can refill the balance.

ThaiGlow uses TypeScript, React, Mantine, and an Express backend, with AI models connected through OpenRouter. The current image workflow includes Grok, Seedream, and FLUX options, while Seedance generates the short videos. SQLite stores the credit ledger, and persistent server storage holds the media and queues.

It remains a private personal project, built around one person's real creative routine.

Release history

Version 0.8.20