My fave blob to date! this was all about experimenting with the spring SOP and external forces, but what i am the happiest with is the material - the highlights turned out great. I still dont really know how the spring SOP works... that was all guesswork... but it looks pretty damn cool imo
sooo this was an attempt to just make a full surreal scene, inspired by @petertarka and @madebystudioq on IG, though i departed from their brighter pastel palette.
this was the most I’ve used premade objects - the head and tree - stock models are not easy to come by, but i could see building a nice library of assets up.
this is also my first ‘scenic’ piece - first bits of architecture / creation of space. Im amazed what ends up taking the most time during these. the staircase came together in about 15 minutes, the archway took me an hour and a half to figure out.
the other big first was incorporating PBR texture - in the checkerboard floor and, less noticeably, the stucco pattern on the walls. Ive got a lot to learn still - especially in how to scale and light them, but theres a lot of potential there.
the window glare and eyes are made bloom which was pretty easy to set up. there’s a nice depth of field effect that is a first for me, but helps with the drama. i tried to create some dust in the space with a particle system, but... didnt quite work as id hoped, and i couldnt figure out how to get it to respond to the same forces as the curtain, which would have been cool.
It kind of became a throw everything at the wall project, one which eventually just needed to be wrapped up. Ill be revisiting these concepts, but the first try isnt going to be a masterpiece
a two-parter in terms of new challenges - cloth and iridescent textures
the cloth sim is really cool and these came out exactly how i hoped. after a few hours of trying and failing to put something together that combined a rigid body world AND cloth sim, I realized there’s a lot more to be dug into here.
i am absolutely obsessed with the cloth texture though - and definitely will be using it again. it was my deepest dive into material notes and ive got a better grip on what im doing there.
the cloth itself ended up being kind of low poly - partially just to save on render time, but the striations look pretty cool
also the lighting and backgrounds here i think are some of my best yet.
a first experiment with rigid bodies! the physics itself wasnt actually that challenging - lining everything up took a bit of work, and its a bit hard to predict without running the sim, but it all made enough sense.
the biggest challenges / improvements were in the obstacle modeling - both the macaroni and cone have thickness that took some working on.
materials and lighting, camera work here were getting better, but still leaves something to be desired. ive almost run out of physics engines to try out, so that will be the next step :)
ok heres a lil thing i built in TouchDesigner using the particle sim and some simple geos as surface attractors - this was actually more intuitive of a project than i anticipated (nice for a change)
I put these together over the past few days and super stoked about how they came out
The process started with the 2nd gif, and trying to model the bumpy corona virus image we see all the time. After a frustrating bit of trying to model it using arrays, i found a simple enough youtube tutorial and just modeled it from a singe UV sphere.
I started to play around with the fluids afterwards and got a somewhat cool sequence, but i rendered it out at too low a res at the end of the day... more on that later.
At the start of my next day, I decided to update Blender from 2.7 to 2.83 at the start of my next working day. Not my smartest move.
Turns out, they basically overhauled the fluid sim during those releases, and on top of learning a new UX, had to relearn how the fluids work.
Which brings us to the first gif, the one that looks like a Listerine ad, and was created to fumble my way around the updates.
It seems to me that the old system was great at creating static water that gets disrupted by an object, while mantaflow (the new process) is way better at creating moving fluids that interact with objects. Im sure this boils down to user error, but thats my first impression. I tried to update the corona splash scene, but i couldn't get anything more than a ripple or two, not the volatile splash i had seen with the old version.
I also think the new process is far more powerful overall, and I like the results the best out of the bunch.
I then learned I can still open the old versions of blender, and was able to do some work on the corona splash one, including a higher quality output.
Felling like i had a good handle on liquids hitting things, i wanted to try my hand at filling something, which brought us to gif 3, which was sort of a funny accident. I still have some things to learn on modeling, like how to fuse together the sphere and column geometry so the fluid sim doesnt spew out the seam. I also find the fluid response to be wildly disproportionate and inaccurate (why didnt it back up the column?). The fluid itself seems to be more blobby, and looks a bit like smoke inside the sphere, but i thought it was funny so why not.
I was excited to try eevee in 2.8, as people talk about its speed. While it was fast, my renders looked so flat and lacking shadow that i stuck with cycles. I think it boils down to needing to get better with my materials (though im proud i was able to make a transparent bluish fluid without tutorials, and the face groups on the virus!).
I also had my first introduction to working with nodes in blender, which while its familiar from TD, offers such another deep well to dive down.
And i am still struggling to find a way to render that doesn't tie up my computer for entire days. Each of these was exported differently, as i figure out what works:
Listerine - lowered frame count to 12fps, set lower encode quality. The frame rate difference is noticeable for sure. I also did this on my old work laptop which has a bit more juice, and i think it helped.
Corona beach - i think i dropped the encode quality and dropped the resolution to 50%. I tried 25%, but it was pretty gross.
Spewing - full res, medium encode quality, but what i think helped here was just keeping it short and sweet at 50 frames.
So, moving forward, i will likely aim for quicker clips, use the work comp, and keep aiming for that sweet-spot of speed and quality.
so i wanted to play around with some grass and wind physics, and it ended up consuming my laptop.
the process for the grass and wind itself was pretty easy, and guided by a youtube tut.
I played around with the topology a bit to make it more interesting, and im not sure if the balding patches are from something in the particle settings, or a result of it not being on a flat plane.
I was running into render time issues pretty early on, but im patient. the real learning moment came during rendering out the animation. It took 36 HOURS to render 9 seconds!
Ive said this before, but as im wading deeper into this pool its becoming a more pressing matter, if i want to keep doing this, i need a better machine. I cant justify a new computer for just a hobby right now. Itll have to be all in or nothing. we’ll see.
took a break from the coursework to play around in AE a bit more - missed doing aesthetic stuff a bit and wanted to make something purdy
some learning moments in here regardless - noise / grain, maintaining underlying transparency, general ‘drawing’ stuff
in hindsight... the clouds look really low res, the planters shouldnt have skewed with the rest of the plant, and the sun got distorted when i resized the ‘inner door’, but w/e... all fixable
style inspo coming from @georgegreaves_ on insta.... check out his work, its beautiful.
Ok - fell behind a bit on this... not because i havent been working on it! ive been knee deep in AE the past week or two working on a secret project for my band thats been taking up a lot of time.
These are all from a week or two ago and are some experiments with textures in blender - all of these feel fairly crude, but its a bit step forward in modeling (look at that glass! look at that.. amp head?). It all still feels so crude, but its coming along one day at a time.
I’ve been sticking with the blender course over the weekend and am about 2/3rds through it - We’ve covered basics on textures, materials, lights, world, UV mapping and projection mapping... honestly im going to have to double back on a lot of specifics, but its helpful to know the words for when i need to double back
The point is, it hasn’t made for thrilling updates so far - here’s a couple of steps along the way
Started on some Blender tutorials because i kept running into modeling limitations in C4D lite - this was supposed to be way cuter but it turned out kinda terrifying
Started to dive deeper into C4D, and found an amazing course on Lynda thats helped de-mystify a lot of the interface. This was my first ‘imagine and execute’ project in C4D and feels like a good starting spot.
Lets see - learned how to keyframe in c4d, learned how a spline works, used extrude and sweep with a helix and a circle and this is what we got.
Look at this lil kooshball lookin thing! I love it!
before hopping in - the quality is brutal - this took my poor lil comp to the limits and I had to do everything in my power to try and capture its true majesty, and the resolution took a pretty big hit... more on that later.
So this is a fun evolving texture thingy that i built following instructions from THIS tutorial - and I really like this guy’s style.
Let me see if i can summarize:
Build one instance of the capped tube, instance it over a massive grid.
while thats cooking, prepare your randoms. start with some raster noise, sized to the resolution of your grid. make it evolve a bit. Convert this into channel data and Slice RGB into separate channels - these control each feelies rotation, color, and z axis rotation. Feel the color channel a gradient ramp to move around in. Add an LFO to control the camera pan.
...Thats I think as detailed as I can get - there are still some pieces of TD that truly confound me. But it also illuminated a ton on the way.
First off, Instancing is awesome, but it eats up processing power quickly. I had to install a slider attached to change my grid resolution just to be able to make adjustments and keep double digit FPS.
Secondly, I learned that if I turn off all of the object displays in the TD environment, I can double my FPS. If I move in the environment until all objects are off the screen, I can gain 2-3 extra FPS. Still, at half resolution I could only render at about 20fps. I will need a better machine if I keep going along this route.
Finally, I learned some Python, which is so handy in TD for tying parameters together. neato.
20 YouTube tutorials later... I finally cracked the audio responsiveness! It’s an evolution on the starting blob a post or two back, and adds some particles into the mix. The whole thing is overlaid on some stock footage to make things a bit more dynamic.
The patch takes a live audio signal and chops it into 3 channels based on incoming frequency.
The Lows blob the sphere.
Mids extend the edges on the particles and blend into noise when pushed.
Highs do something to the particles that... I kind of lost track of. It was supposed to be a camera move but Im not sure if i have it set up correctly.
Im really excited about this route, and live generative visuals are something I’ve dreamed of for years.
The patch could use some tweaking for sure, but there’s really only one thing standing in my way of actually implementing this in a live environment: the latency is KILLING ME - its like a solid beat behind the live signal. Probably more a result of my now 10 year old audio interface, but its just another piece of gear to add to the list.
Speaking of which - the incoming audio is being sequenced live by me on a Korg Monologue, Volca Beats, and Volca bass.
My partner is a 2nd grade teacher and needing to adapt to teaching remotely. I put together this hold screen in After Effects for her for when she needs to step away.
It started with just lines and shapes, but they started gaining some meaning as I went. There’s a little city scene in there, and portrait of sorts.
Its gotten pretty good use so far, and the kids think its cute, but are divided on the portrait...
Devin Castaldi-Micca @devincastaldimicca - Tumblr Blog | Tumgag