You can group select all of the NEW keys the camera are suppose to āteleportā and set this blending
Normal behaviour is free and this means it will try to interpolate the camera between the two keys and it wonāt produce a clean cut
This is true for unity. Currently in needle i believe it is interpolated. You can notice it in the demo with a bit of a wonky cut. If you would go frame by frame you would see that it actually is trying to move towards to the new shot rather then just appearing in that state.
From what i understand this might behave the same in the next updates. Not sure. Would be lovely.
.
.
After a working animationā¦ i wanted to HIDE and SHOW the two UIs based on where the animation was. Since the car REAR UI was visible in the initial shot.
I never used timeline before so it was a pleasant experience since it was easy to use and needle supported it fully
.
.
.
To wrap this up,
The visuals of this demo are far form perfect. Many details look quite flat or dark. Some visuals were different in unity and web. Some were the same
I just realized i didnāt event try using post processing :facepalm: hehe, well, maybe i could have used the ACES tonemapping to my advantage. Well, good to know for the next time.
Wow, thanks for that excellent write-up!
Regarding gradients in shaders, you can actually use Needleās ShaderGraph Markdown package to add support for gradients in the material inspector directly.
Regarding constant animation, we noticed that issue too and are looking into ways to make it better - as explanation, while Unity allows to specify interpolation per keyframe, threejs only does it per entire track, so we need to bake constant values into ālinear interpolation in near-zero time differenceā.
Thanks a lot for that great summary of your process!
Awesome writeup, thank you! And happy to hear the timeline worked well for you i think initially you wanted to go more into the configurator direction right? Was there anything that blocked you from doing it or just shift in interest/time/other factors/me remembering it wrong?
Pleasure was mine!
I worked on this demo since the public release, couple hours a week. The progress was slow. So a lot of my decisions were made earlier.
One aspect i couldnāt really overcome was the need for screen space UI. I have practically 0 experience with web dev and with styling.
One thing i was VERY confused about was the shaders and materials in regards to Three.js and how much needle actually does. I could have just carefully read the docs but noo, i had to try it for myself. For example i took the PBRGraph and tried to edit it ā¦ and ā¦ yeahā¦ it took me a while to realize that itās just there to represent how it will look like in unity and that it doesnāt define anythingā¦
Configurator is definitely possible If i would pour more time into it it would be achievable
My actual goal was just to research and make a proof of concept. To decide if Needle was capable enough that it would let a Unity developer with slight understanding of three.js and slim to none typescript experience make an interactive experience. Secondly, how well the content is converted since we currently have an overhead exactly on that. We have to setup and maintain the content in two frameworks.
My overall experience hints that all of the above seems to be possible. Thatās mainly because you seem to be planning to include more and more features.
But these are the 4 features i lack the most.
- Screen space UI support
- Custom Lit shader support
- URP Lit with specular workflow support
- Full post processing support
I know that ā¦ all of them are enormous chunks of work and that making these have to make sense.
I understand that Needleās general direction more implies the need only for 3D UI.
I understand that Custom Lit shaders are hell to support well. (again, even with the specular workflow- which is the least probable to happen)
I understand that URP Lit with specular workflow is possible but not exactly a high priority
I understand that Post processing is planned and SOME will be available ||When i was trying to migrate the three.js example https://threejs.org/examples/webgl_postprocessing_unreal_bloom.html i kindda wasnāt sure what depended on what, these examples are greatā¦ but are too tied together ||
I hope i will get to the example again and add more features in
Thanks again for the detailed explanation and info. Really valuable feedback as always
Fantastic tech breakdown of the demo, really insightful to see how you dealt with lighting
by user 103054507105067008
Thank you That would be all from my side.
Any notable feedback you guys want to give me in return?
So far i got ā¦
- camera movement is too slow
- floor could be reflective (SSR? )
- Lens flares - shining lights
- Wheel are spinning too slow in regards to the grid which is moving too fast
I didnāt read every comment.
It looks really nice.
But what is the use of THREE here, if the user doesnāt have the possibility to interact with the camera/model? Why not just prerender a video in Blender and use it?
by user 367081956959322114
Oh yeah! Most definitly. Letās say that a nice orbital camera is an easy task. And potentially carpaint selection and wheel selection would be present. A basic Car configurator. Such as: https://playcanv.as/p/RqJJ9oU9/
I suppose my question is mainly focused on visuals then :Emoji_Sparkles: