- What is the type of project? modeling, rendering, motion graphics, games, VFX etc
Primarily modeling, with focus on allowing interaction with Python later so that we can script moving the camera and generating a large dataset of RGB and depth info. We already know how to do the camera motion, rendering, etc.
- Are you a company or individual? We recommend to post a link to your website.
We are a university research lab. You can see our work at https://labs.wsu.edu/m3robotics/. We currently have a research project for optimizing the plane de-icing process through the use of computer vision, robotics, and machine learning.
- What kind of designer are you looking for? Experience level, special skills, etc
We have no preference on skill level. We just need the solution to meet specifications.
- What is the deadline?
15-30 days
- Description of the project. It is best to be clear on what you expect the final project to look like, this will attract higher level designers
In the early stages of this research project (before winter comes and we can go start collecting real data from airplanes at the airport), we want to generate a training set for our machine learning exploration that can simulate ice accumulating on the top surface of wings and fuselage.
We have tried to muck around with UV unwrapping and textures with the BSDF nodes editor to create rough surfaces. We have been able to make it kindof work, but are having real problems being able to specify which regions of the wing surface and fuselage have the “bumped” texture and also control the color of the snow/ice.
So, we are looking for a Blender file where we can load two different files:
(a) a mask image that tells where there should be snow/ice and where there should not. 0 means no ice. 1 means maximum height. We can make a noisy mask to have the snow/ice be rough.
(b) a mask image that tells what the color of that snow/ice should be
Ideally these masks would look something like the UV unwrap, where it is easy to identify which regions of the mask images correspond to the top surfaces of the wings and fuselage.
The idea is that we can later create a Python script that
(i) programmatically generates the two mask images
(ii) applies the mask images to plane model to regenerate the snow/ice+color
(iii) moves the camera to a random location, but still pointing at the plane/wing
(iv) save this render out to disk as an RGB image and a depth image and the corresponding mask regions as seen in the camera view
We really only need step (b) above. We have done these other tasks on previous project and can adapt as needed for this project. Ideally, the deliverable would be a model that can take in the two mask images and generate a render with the snow/ice and desired color(s) added.
- What is the budget? This is not obligatory, but it helps a little information, like whether this is for entry level or veteran designer.
We do have funds for this budgeted into the research grant. I suspect that someone who understands Blender better than us could get this done pretty quickly. We almost have it working, but can’t figure out why certain things don’t change when we try to programmatically update textures.