Suggestion: Expanding the exercise image database

[seivanesandro]

Use case

Dear wger Team,

First of all, I would like to thank you for the amazing work you are doing with the wger API. It is an incredible resource for the fitness community and developers alike.

I am reaching out to suggest an enhancement regarding the exercise images. While the current database is very useful, I believe that adding more visual variations or increasing the number of exercises with high-quality illustrations/photos would significantly improve the user experience for those integrating your API.

Having a more comprehensive visual library helps users better understand the form and execution of various movements, making the applications built on wger even more robust.

Are there currently any plans to expand the image assets? Additionally, if there is a specific way the community can contribute with new images that follow your guidelines, please let me know.

Thank you for your time and for maintaining such a great project.

Best regards,

Proposal

No response

Additional Context

No response

[rolandgeider]

hey! Glad you like it! You are right that many exercises are missing images, you don't need to convince me 😄

The problem is that we want the images to be in a free license (CC-BY-SA or similar), which is difficult to come by (and to be fair, most of the ones submitted by the community are probably not, this would need to be cleaned up). We were also thinking of using AI to convert such images to illustrations and have at least that, but that brings other problems. So yeah, there's no silver bullet at the moment. Would you be interested in contributing here?

[seivanesandro]

Hi Roland, Thank you very much for your reply and for agreeing that expanding the image database is important. I think the idea of using AI to convert images into illustrations is a great one and could be a practical solution to the free license challenge. I am interested in contributing. What specific areas could I help with, and what would be the first steps? Thanks, sandro Em sex., 6 de fev. de 2026 às 13:15, Roland Geider ***@***.***> escreveu:

…

*rolandgeider* left a comment (wger-project/wger#2196) <#2196 (comment)> hey! Glad you like it! You are right that many exercises are missing images, you don't need to convince me 😄 The problem is that we want the images to be in a free license (CC-BY-SA or similar), which is difficult to come by (and to be fair, most of the ones submitted by the community are probably not, this would need to be cleaned up). We were also thinking of using AI to convert such images to illustrations and have at least that, but that brings other problems. So yeah, there's no silver bullet at the moment. Would you be interested in contributing here? — Reply to this email directly, view it on GitHub <#2196 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ASEKVCXUTZQCYNT72U75XDT4KSHXFAVCNFSM6AAAAACUDQKZUGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTQNRQGM4TANJYG4> . You are receiving this because you authored the thread.Message ID: ***@***.***>

-- Sandro Seivane

[rolandgeider]

awesome! Do you have discord? That's probably the easiest and fastest way to coordinate ourselves. If not, we can just discuss everything in the issue

[seivanesandro]

Hi Roland, Thanks for the quick follow-up. Unfortunately, I don't currently have Discord. I'm happy to continue the discussion and coordinate on the GitHub issue, as you suggested. Best regards, Em sex., 6 de fev. de 2026 às 15:33, Roland Geider ***@***.***> escreveu:

…

*rolandgeider* left a comment (wger-project/wger#2196) <#2196 (comment)> awesome! Do you have discord? That's probably the easiest and fastest way to coordinate ourselves. If not, we can just discuss everything in the issue — Reply to this email directly, view it on GitHub <#2196 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ASEKVCWVY7QLKMLDNZ2KL534KSX4RAVCNFSM6AAAAACUDQKZUGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTQNRRGA3TSOBQHA> . You are receiving this because you authored the thread.Message ID: ***@***.***>

-- Sandro Seivane

[rolandgeider]

No problem!

Here are some ideas / thoughts:

Cleanup the exercises:

• There are lots of duplicated exercises. We would need to find them and combine them. This is mainly due to some people only knowing them under different names (there’s the „alias“ field for that, but not everybody searches before submitting)
• Make sure the English description is actually in English. We recently added a check so that this is the case, but there’s still many exercises where this is not the case.

Cleanup the images:

• Make sure the images are not GIFs or similar. These can’t be uploaded anymore, but existing ones still have them.
• Make sure the images are actually CC-BY-SA (a simple reverse search should be enough). If not, we should remove them.
• If we decide to go the AI route, start converting them. We can mark images as „foto“, „sketch“, etc. Perhaps also add an „ai“ toggle? Alternatively, start looking for actual pictures from Wikipedia or other free sources (we do have a UI to add all the source metadata when creating an exercise, but I don’t think we use it when editing. Would need to check that), every bit counts.

For some of these (like editing the description) you don’t need any special rights, just an account older than 3 weeks, but I would need to do others like combining the exercises or deleting images.

[stuarta0]

If someone is willing to clean up the duplicate exercises, I could consider creating a new set of standard open license exercise images/videos. I would use a 3d rendered character model, with optical motion tracking to animate the character rig. I could release the blender file with character, rig, and lighting and rendering setup if I'm not around to expand the exercise database in future.

Is this the kind of thing the project would be looking for?

If so, I'd need some feedback: 1) an initial workshopping of the render style and animation, and 2) someone to review the images/videos for correct depiction/technique.

Edit: a future proposal could be a 3d viewer to allow the user to view the exercise being performed from different angles 💪