WorldKit : suite à l’échec du Kickstarter, l’outil de terrains se tourne vers l’open-source


En janvier dernier, nous vous présentions WorldKit, un projet d’outil de création procédurale de terrains. Une campagne de financement avait été lancée sur Kickstarter.
L’engouement n’a manifestement pas été au rendez-vous : avec seulement 24 contributeurs et 1000 euros engagés, la campagne est un échec.

Adam Brown, le développeur principal, vient cependant d’annoncer qu’il ne déposait pas les armes. Dans un message posté sur Youtube et que nous reproduisons ci-dessous, il explique désormais vouloir suivre l’exemple de Blender et opter pour une approche open source avec financement par la communauté.
Il précise que les retours lui ont permis d’ajuster le positionnement de l’outil, avec un projet au workflow plus complet et des améliorations, entre autres, sur le viewport. Il souligne aussi avoir reçu un accueil très différent entre les communautés Unity et Unreal Engine, cette dernière étant bien plus réceptive.

Une nouvelle démo vidéo devrait être proposée d’ici quelques semaines, suivie d’une nouvelle levée de fonds plus longue que la campagne Kickstarter. Nous aurons donc l’occasion de revenir sur ce projet.

I am still working on WorldKit as a solo project, so it’s going a bit slower. Because I failed to get backing to support a commercial release of WorldKit, I am planning to follow Blender’s example and raise money to pay for the current development to this point and then open-source it so the community can support it.

One of my takeaways from the initial campaign is that WorldKit needs to be able to replace current tools like World Machine, World Creator and Houdini completely for the whole world generation process, not just the first step in the pipeline. Because of this I’m working on improving the high detail features so it can generate 16k maps in less than 30 seconds with a much higher degree of realism. The new high-detail system uses AI that is trained to reproduce landscape features from satellite imagery which allows for very efficient generation of terrain that is ready to go straight to engine without the need for other software in between to achieve high levels of detail.

Another thing I received lots of feedback on was that I needed to improve viewport controls and visuals inside WorldKit. To this end I have added a « preview » mode which generates similar in-viewport visuals to the video above in about 2 seconds. I also added full detail mode which takes about 20-35 seconds to generate a much higher resolution landscape in the viewport, so that it’s easy to see all of the subtle detail added by the new AI terrain feature system. And finally, I added more familiar camera controls to the viewport that mimic UE4 Editor viewport controls, and I added a PBR shading pipeline to the WorldKit viewport based on the UE4 rendering pipeline as described in Epic’s paper on the subject. I’m currently working on getting proper skybox, water, and fog shaders into the viewport based on the new PBR rendering pipe.

I received a lot of negative feedback from the Unity community as a whole during the last campaign, leading me to believe that either people are not making the type of game that needs this kind of tool in Unity, or there are tools built into Unity that people would rather use than WorldKit. Conversely, I received a ton of positive feedback from the UE4 community, and it seems the demand for such a tool is much higher there. Because of this I’m focusing on support for UE4 right now in addition to the general features I listed above. I’m hoping to secure some funding to develop a UE4 plugin to import WorldKit worlds directly into UE4 Editor.

I hope to have a video demonstrating all of the new features up sometime in the next two or three weeks. I plan to run a longer publicity campaign prior to the next fund-raising round this time, as I was not able to reach enough people during the 30 day Kickstarter the last time.

A Lire également