well i am not really sure, what if with the animation the side will be more visible , i dont know the plans yet but we need to keep in mind that if there is an animation around the house and spaces that it will also look good from diffrent angles.
That is a good question.
I'm working under the assumption that the camera will stay somewhat close to the house, and low to the ground. No birds-eye view or anything telephoto. This technique would not work under those circumstances.
Homework Week 1 - BG Bushes
@theluthier
I'm not exactly sure what to do for this first assignment, since these bushes are so far away.
And it's hard to get much info about "L" from this picture.
So I've been focusing on testing out different techniques with respect to efficiency and accuracy to the artwork.
As you can see above, I think I'll be using the Ghibli method to make "clusters" then rendering those to PNGs, and using a few planes to build out the bushes. Here's how it looks:
From here, I can keep refining the shader to match the various colors. And animating a little sway.
And I think we can use these cheap bushes to fill in any gaps in the background as we figure out exactly what will be surrounding the house.
A few questions moving forward:
Will the camera stay close to the house, and low to the ground. These planes will look mega bad from other angles, but I think the cheat will work if the camera is somewhat constricted. Within the fence at least?
Can we make a shadeless material in Eevee?
Should we?
I see emission, but I don't want the bushes to cast light obviously.
Also, I considered making them not cast shadows, but just adding a simple ground shadow manually. So the planes won't shadow each other and break the illusion.
P.s. I wouldn't say you are a brutal grader, but you do always give constructive criticism. It's the only way to grow, and just hearing "great job" doesn't really help with that.
I'm sure the camera is going to be mainly around the house, the distant objects are not really a concern as far as camera angles are concerned, and when the time comes a constraint can always be added so the plane always faces the camera.
I was thinking a constraint on the whole bush object could work to subtly shift it toward the camera.
If individual clusters were constrained using rotation, they would end up colliding.
I would look at the image and suggest 'L' is actually part of 'M'
The Orange and Brown bushes are the Mid ground, and the greenie bluey color is the Background bushes or trees.
What do you think? ppfbourassa and @theluthier
Being so far from the camera I'm not sure they would rotate that much, and as long as one is slightly behind the other.
Can we make a shadeless material in Eevee?
Well, there is the Shader to RGB Node:
Have a play with that ;)
If you see the Ghibli Clouds tutorial by Lightning Boy he uses this method with a Shader to RGB node
I used Shader to RGB to get to this point, haha.
I mean, in Blender Internal you could set a material to "shadeless" and it would just not factor in any lighting or shadowing whatsoever. A red material would just be #FF0000 all the way across the surface.
And for the extreme background, that could work because the angle won't really change. If we decide to change the lighting to night or whatever it wouldn't work anymore. Just a thought since I'm "baking" the lighting anyway. And I'll have to color-correct the PNGs to counteract the scene lighting, rather than just getting it right in the first place. If that makes sense.
And I agree Adrian, "L" might not be real.
ppfbourassa You might want to grow your 'L' Bushes into 'M' Trees over the next week, lots of watering required.
ppfbourassa Crap...I was going to say "great job"... 😉
You and tijnkroon absolutely crushed this first week with minimal guidance from me. I LOVE that about this community. It's the kind of resourcefulness I hoped - nay - knew was be there. For the quality of result and tests you get full points + some extra 👏
Will the camera stay close to the house, and low to the ground?
I don't plan to restrict the camera in any way. But I definitely want to prioritize quality / optimization based on distance from the house. I will explain this during the stream tomorrow, but FG assets (closest to the house) will be max fidelity, MG will be optimized a bit more to require less resources, BG will be max optimization.
The expectation will be that viewers stay closest to the house. This means that the environment only needs to look good from within the FG zone. But viewers won't be restricted to navigating into the BG area (for seeing how its made for example).
I think your card system is likely the way we'll need to go for BG assets. Now that you've run a test for the real geo and it's not looking good ha. Once I introduce lighting (week 3 I expect) we can lock it and your bakes will look natural. Only note about your current bake is I'd like to see more gradient from light to dark:
Glad to see the "shadeless" advice given. We'll no doubt good plenty of good use out of that technique.
For week 2 I'll tell you the same thing I told tijn: For week 2 it'd be nice to get a couple more variants and test some population techniques. You can test on a randomized ground plane or grab the terrain meshes I'm working on. Very curious to find out how optimal these can be once populated.
Great work ppfbourassa with bonus points from @thetoughcookie 🤣
@theluthier , we have a question about BG Bush 'L' or maybe the lack of.
Is Bush Cluster 'L' a bush or part of Tree Cluster 'M' ?
The image you mention above is of MG Bush Cluster 'K',
I'm asking because I don't want to lose Parker, a valued team member here in the wilderness of the background to our rivals in the foreground.
Good question.
My current plan is to keep working on low-poly tech, and just prepare to fill in the gaps when the full 360 needs to look good.
But I'm open to other tasks if needed.
@adrian2301 Wow that "L" is confusing isn't it...I think I was ambiguously pointing to where BG bush clusters would be. While I don't see them explicitly in the art, I know they will need to be there to correspond to their FG counterpart.
My current plan is to keep working on low-poly tech, and just prepare to fill in the gaps when the full 360 needs to look good.
ppfbourassa Perfect plan!
Today I learned about a workaround to a (maybe) bug in Eevee with image planes.
If I use the "import images as planes" addon, the planes pop in front of each other using the object origins, rather than intersecting in the way that would be expected. If I set the Blend Mode to Alpha Clip rather than Alpha Blend this doesn't happen anymore. So that will save me a headache.
Also, I set up a gradient control in the material. I figure the PNGs can be used for the luma only, and hue-shifted for different bushes like green and blue-ish ones. And then the gradient can also be tweaked to give more variation.
Old, hyper-pink bush shown for reference.
Setting the material to a Roughness of 1 prevents the "glossy photo" look, since the lighting is already included in the PNG.
My next goal is to make a few more bushes that ready to be instanced around using (maybe) a particle system, or geometry nodes.
I feel that a hand-places approached might look better, and not take up too much time, but I'd like to try out geo nodes so I dunno.
Also, I might open up photoshop and give some the "fire-like" tendrils shown in the artwork.
I'm not "in love" with this look, and I'm afraid it might actually look like fire if I'm not careful, but I'll try it out.
Homework Week 2 - BG Bushes
@theluthier
See the post above for images and updates.
I uploaded to /SCENES/MODELS/TEMP because these don't fit into the "placement file" workflow (or at least I couldn't find mine.)
The gradient was a good call for sure. I'd like to bake them into the PNG at some point, but for now it's flexible.
These still intersect with the ground in an ugly way, and I'm not sure how to render them "always in front" or anything like that. Maybe if I fake a contact shadow, they can just float above the ground so the don't intersect? Or maybe the grass will hide the seam.