What's new in Release 19.5

June 2026 Enhancements (v19.5)

This DBGallery release enhances video AI with scene-level summaries and custom prompts for video, updates the login page style, and improves error messages in share links. Additionally, it upgrades the non-video image AI model to gpt-5-mini for more accurate, less restrictive descriptions.

Get more from your video content

We've made video content significantly easier to find and discover. With our new video AI timeline and description features, scene analysis (aka scene descriptions) is automatically generated during pauses in speech, or if there is no speech in the video, rather than relying solely on spoken text. Those non-spoken scene descriptions are in square brackets in both our transcript text and the timeline.

Release 19.5 video transcript timeline

Fixed Video AI Descriptions: Resolved an issue where video descriptions could occasionally return blank results or false positives indicating potentially harmful content.
Custom Video Prompts: Added custom prompt capabilities for video files, giving you total control over the AI-generated summaries you want. This mirrors our existing custom prompt feature for standard image files.

Login page style refresh

Enjoy a fresh layout that is cleaner and easier on the eyes! There have been no functional changes—only visual upgrades. We've also simplified the sign-in field label from "Email Address or Username" to a cleaner "Username or email" to make logging in intuitive right away.

Release 19.5 login page style refresh

Sharing Improvements

Improved handling of non-existent or deleted share links.

We know how crucial seamless file sharing is, especially when external stakeholders, clients, or sponsors are viewing your work. To ensure a professional experience, we have eliminated generic "Server error" pages. Instead, users will see a clear, helpful message indicating if a link has been deleted or pasted incorrectly. We also now provide specific messaging if the shared images or underlying folders have been removed.

AI Upgrade

For standard images and photos, we have upgraded our core model from gpt-4o-mini to gpt-5-mini. This shift brings vastly improved custom prompt capabilities and richer descriptions. Moving to OpenAI's native pipeline drastically reduces overly sensitive moderation filters (such as previous limitations with Microsoft Azure), allowing you to seamlessly index and describe common everyday scenarios—like photos of family pets at the vet—without interruption.