monoCAMERAl
This game is from the old Ludum Dare platform. You can try viewing the original page on Web Archive
(may not load correctly).
(may not load correctly).
You've been transformed into an imp, and trapped in a camera!
Now it's up to you to draw your way out by capturing the evil warlock Moorcock's vacation memories!
Stick it out, and you might be rewarded!
- Art + Concept: Jenna Fizel
- Programming: Russell McClellan + Open Source Contributors
- Scenario: Russell McClellan
With apologies to the late, great Terry Pratchett.
On a personal note, this is the first LD game jam I've entered since LD4: (http://hamumu.com/LD48/ld_entrants.php?user=logicalhippo)!
Most of the time was spent on the image processing. No neural networks were harmed in the production of this game. We ended up going with the image similarity metric described in http://web.stanford.edu/class/cs231a/prev_projects/final_report.pdf (Bo Zhu + Ed Quigley's CS231a project from Stanford), using local features from the basic HOG algorithm implemented by harthur (https://github.com/harthur/hog-descriptor). Our codebook was trained by k-means on 20,000 sketches from the dataset by Eintz., et al from here: http://cybertron.cg.tu-berlin.de/eitz/projects/classifysketch/.
Vacation photos provided from the sketch retrieval benchmark dataset by Eintz., et al available here: http://cybertron.cg.tu-berlin.de/eitz/tvcg_benchmark/index.html
(ugly) source available here: http://github.com/russellmcc/ld37-imp
Thanks for the time and attention!
Now it's up to you to draw your way out by capturing the evil warlock Moorcock's vacation memories!
Stick it out, and you might be rewarded!
- Art + Concept: Jenna Fizel
- Programming: Russell McClellan + Open Source Contributors
- Scenario: Russell McClellan
With apologies to the late, great Terry Pratchett.
Technical Deets:
On a personal note, this is the first LD game jam I've entered since LD4: (http://hamumu.com/LD48/ld_entrants.php?user=logicalhippo)!
Most of the time was spent on the image processing. No neural networks were harmed in the production of this game. We ended up going with the image similarity metric described in http://web.stanford.edu/class/cs231a/prev_projects/final_report.pdf (Bo Zhu + Ed Quigley's CS231a project from Stanford), using local features from the basic HOG algorithm implemented by harthur (https://github.com/harthur/hog-descriptor). Our codebook was trained by k-means on 20,000 sketches from the dataset by Eintz., et al from here: http://cybertron.cg.tu-berlin.de/eitz/projects/classifysketch/.
Vacation photos provided from the sketch retrieval benchmark dataset by Eintz., et al available here: http://cybertron.cg.tu-berlin.de/eitz/tvcg_benchmark/index.html
(ugly) source available here: http://github.com/russellmcc/ld37-imp
Thanks for the time and attention!