Update - Week 51 - 2021
Nearing the end of the year means one last push to finish expectation - including testing a city layout and refining the "shower door" issue.
January 23, 2022
Weekly Sprint Stats:
Points Velocity per day
This week was about pushing hard to try get a final animation out. Though client is happy, I am way behind my milestone schedule. It has been getting me down, but I did do a tonne of hard work to get things going. Again, I focused mostly on the painterly project, including developing the post processing and doing animation improvements - making the paint come alive. This also included testing the problematic "shower door" effect with the animation systems, proving to not quite be effective. At the end of the week, I got the eureka moment and decided to start prototyping the "sunny rain" 3D post processing idea. It worked. Now I have one more week to work on the other client that has been on hold to close the year on a good note - and holidays for two weeks!
These updates, yes, are a little late, but I hope to get to the present day by publishing two updates a week for the next couple of weeks. I have also had a couple of holidays over new years which were much deserved and long time coming! I haven't had a 2 week holiday in more than 2 years. Time to get away from the computer.
Again, I hit my work average, despite it being Christmas week. I made good time even with the social meetings and such - and through the distracted mediums I managed to find effort. I still am working under capacity, which is still discouraging. I hope next year I will get to a point where I have production capacity and I can start my work again anew - with the right foot this time! When my sprint efficiency hits more than 100%, I will be happy to say I can take on more work, hire people, and actually register my business as a company, till then I will remain a freelancer for now and work within my means. Though, on a good note, this week did have a high point to hour ratio of productivity - which hit 7.2 points per day. This is excellent! Nearly double from last week. This means.. I did more of what I had planned given the days available to work. It's always good to look back on this, reminding myself what a great Christmas season I had. Family, food, donations, children, drinking, eating, and all round festivities - AND good art and exploration to a productive high. It was a good week.
The Painterly Project Updates
This project got a lot of love this week, including the following:
- New Billboards
- Stress test for using geometry nodes to make a layout (ultimately it would crash though)
- Updates to the "shower door" post-processing technique
- New animation systems to make the paint come alive with higher frame rates and frame blending - instead of the not desired and non-temporal coherent frame jitter.
- Bug fixes when it comes to offsetting normals large distances
- Model updates for the Launch Pad and Workshop
The week started with me working on making billboards and trying out my generative city layout tools. I could quickly make the layout, but I unfortunately found issues with performance and even file corruption. I will have to find a way to easily freeze my generative city chunks and use the frozen pieces for layout instead of the live geometry nodes trees. This makes life difficult, freezing geometry nodes is no joke - I may just have to learn to export to FBX and import again, pray to the gods that the vertex colours get preserved. I have yet to run that test.
On a side note though, I did find a way to make frames offset, then blend, then offset, then blend again. Despite what many artists have shown me as reference and preference with the general tendency is to mimic real stop-motion paint based frame blending, where you capture one frame, then paint over, capture the next frame, paint over - I couldn't buy it. This makes a very jittery experience and for temporal coherence, it doesn't allow the brain to blend movement very well breaking the illusion that this is a story we are supposed to submerge into. I don't have a lot of data to prove my case, and my control group and test group to prove the theory is extremely small, but those outside the art circle most definitely prefered higher frame rates with less frame jitter - without a doubt. So - one of the goals with the client was to make this somewhat trippy - and to do that I will do temporal coherance with actual brush blending on higher frame rates - more to simulate how our eyes see (temporal coherance a bit like dreaming AI) instead of shutter speed from mechanical cameras.
Tests proved relatively successful, with a method that would swap and offset the brush strokes and blend them over, then swap their position again. But.. I still could not get my pre-processor to look good. This had me fight yet another of my known issues: the "Shower Door" effect. When we do post-processing, after we capture the image, the data is taking essentially flat information and processing it. This is a bit like looking through a pane of glass filled with water droplets. You see through the water, distorting the image behind, but anything you move of if you move the camera, the droplets stick to the glass and the world behind it moves in 3D space without them. This can create a 2 layered visual illusion, like you're looking through glass. This "shower door" effect is far from the effect I wanted, or else I would not have developed the shader in 3D space over UV's! So... I needed a solution. After some bug fixing, render tests and more tuning - I ultimately resigned to... an idea I saw on my last weekend holiday trip with the family.
When I was in a small town about 2 hours out of Bogota, Colombia, in the central plaza where we had unloaded to catch our bus - it has started to rain heavily though the setting sun was shining underneath the clouds. Surrounded by green mountains, I could see down the street past the colonial lime washed church down the street to the mountains - and through the pouring rain glistening golden in the light I could see the mountain blur and look... almost painted and abstract as the light distorted through the water droplets. I walked across the street and looked again, I could see the volume of droplets work with the perspective. It was beautiful, and this is my answer.
So, at the end of the week, I recalled that effect as it gestated in my mind - and before finishing for Christmas, I made a quick prototype, where I put the 3D pre-processor (post-processing in viewport with the incoming refraction shader technique on a plane in front of camera to get realtime post-processing from Eevee) into brush strokes basked out into 3D planes in the viewport. This means I could "stick" the brush strokes into 3D space, and as a prototype it worked! This was the last piece of the puzzle to let my mind rest at ease with the paint style - and good timing! End of year, and I have all the answers. Next year I would implement it and then focus exclusively on content, animation and focused work. I will call this new 3D pre-procssing or post-processing technique the "Sunny Rain" technique.
I had no notable progress with the other projects, it was a full week! Similar progress with Bforartists. Looks like everyone is in season, and little progress from Blender since they too are in holidays also keeps things going slow there. But, with Iyad we still slowly but surely have been working the top-level brush switcher panel. Hopefully first trimester next year we will have it implemented as a default in the latest distributions of Bforartists.