Let's continue binge-watching the Blender Conference videos! Here's videos 27 to 45. 

As before, I've added video thumbnails for the videos that stood out to me (in a good way) so it's easy to find the "best" ones.

However, this is entirely subjective and not meant to dismiss the other content! There were many interesting talks, and everyone who was brave enough to get up there to talk is a hero to me.

Your e-mail app or Gmail may cut off the e-mail at some point, but you can click on "View entire message" at the bottom, to load the rest of the content.

Presentation from an architect who does archviz renderings. He uses fSpy and Perspective Plotter. Some other tools he uses are Photographer add-on, Lily Surface Scraper and a few others.

He also uses a few libraries like Chocofur, Poliigon, Sanctus material library. Archviz is like kitbashing, you need many libraries of assets. There still need to be many more people contributing to assets.

It wouldn't be a Blender Conference without an animation talk by Hjalti Hjálmarsson. 

In this case, he goes over the process of previs / layout for a short animated gag. The most important part is not technical, but just the creative process of coming up with different ideas, and deciding what the best version is. 

It's interesting to see the iterative process and how it improves the story and makes jokes work much better. He also seeks out feedback from others during this process, which is not to be underestimated as a valuable tool.

It really shows that there's a massive difference between the first ideas you have and that become a shot list, and then simple sketch-boards / thumbnails, and the final layout after many revisions. 

In his example, it's really a huge quality improvement. So I think keeping the previz and layout process really simple, so that you can be creative and try out different ideas, is essential. If you're a filmmaker, don't miss this honest and insightful talk.

NeRFs and Gaussian splatting are interesting because they have relatively realistic results, comparable to photogrammetry but with the addition of more realistic light reflections, yet at very small filesizes. 

And they work even on smartphones or websites. They can also do thin parts and glass, which is a challenge with photogrammetry.

Besides basing it on real life, the way you would with a camera and photogrammetry, you can also do this for a 3D scene. In that case, you place virtual cameras around your model in Blender.

These students have created a free add-on that makes creating Gaussian Splats from your Blender scene easier.

Reconstructing a lost city, Tenochtitlan, in Blender using height maps, geometry nodes, procedularism. He optimizes things early to make sure there's real-time iteration possible, without it slowing down too much. 

Taking some tricks from gamedev, he uses billboards for trees in the far distance, and instanced trees for closer by the camera detail.

The result is really stunning, you can see the renders here: https://tenochtitlan.thomaskole.nl/

Also, the name Thomas Kole seemed familiar to me. Then I realized, he's the creator of the add-on Seams to Sewing pattern (see also my short video about it) - https://thomaskole.nl/s2s/ 

Flow is an animated movie made in Blender in 6 months as a European co-production from Latvia, Belgium and France. 

He describes their pipeline, and how they communicated with their team. This consists of communicating through Slack and Discord, keeping track of shots in Kitsu and setting up a simple Google website for the styleguide, explanation of software and platforms used in production and so on. Lots of cat references as well.

He mentions some add-ons I've not heard of yet, like Rig Picker and FCurve helper. If you're into animation, I recommend watching this video carefully and taking notes.

One of the challenges on this project was the very long shots with camera movements, 16 seconds per shot on average. 

They're having a lot of success in film festivals: Cannes, Annecy… I look forward to watching it in the movie theatre in January (it depends on the country when it's released, it may already be in theatres where you live).

I just want to mention again: 20 animators made this movie in 6 months, and half of them were Blender beginners and junior animators. That's really amazing.

A talk about a web-based 3D platform that in this case is used by a museum or exhibition space to have a web-based interactive virtual version of their real museum.

An amusing look at how code is rewritten for Blender. Some of the code is 30 years old, from Amiga days… undocumented, nobody remembers why. So he rewrote the VSE (Video Sequence Editor) recently. 

This talk is mostly for programmers, but even as a non-programmer I found it quite interesting.

So many bald men in this video. Does the stress of using Blender cause hair loss?

Inspired by Beeple, Jorge Alejandro Castellanos Paz creates daily renders. He goes through a live creation process together with the audience. I'm curious to see how he can evolve his art if he keeps this up. 

Daniel Martinez Lara shows a geometry nodes setup that takes advantage of the recent integration of GN in Grease Pencil. It allows him to sketch a shape and have it be filled with a blob and material. 

He then talks about an event in Spain for new technologies (Next Lab). He talks about VR use in 3d. Personally, I think VR is really interesting, but having the VR headset on is very tiring, after 30 minutes or so you've really had enough. So for a serious 3D workflow I don't think that's realistic yet. 

But like the speaker, Daniel Martinez Lara, I think the input device (the 3D pen) is the most interesting part. You don't even really need the headset. Apparently Logitech have some kind of new 3D pen - https://www.logitech.com/en-us/products/vr/mx-ink.html (Although it seems you still need a headset, maybe at some point we can use it with AR Glasses!)

The benefit of being able to draw in 3D for animation, is that you get a workflow more like 2D, where you remove some of the typical steps of 3D like rigging. 

If you've ever used the NLA for animation in Blender, you'll know that it can get pretty confusing. There's a great add-on called Animation Layers which I recommend, but even that can't fix some of the underlying issues with the old NLA code. So now they're working on improving this by adding layers and a whole new animation system.

Nathan Vegdahl (Russboy!) explains the concept of Slotted Actions. 

Although it’s a clear explanation, I feel like I have to try it out in practice to really “get” what they’re talking about. Disclaimer, I don’t do a ton of animation.

Basically if I understood the talk correctly, the current/old system has actions for every individual thing. So if an object has a material, mesh data and shape keys that’s already 3 actions for just that one object. Now with slots, it can be just one action, with 3 slots (or however many are needed).

Content driven development! Simon Thommes talks about Blender Studio.

It’s very inspiring what they’re doing.

I’m still not sold on the promise of geometry nodes. I think it’s a tool made for a very small selection of very specific people, something more akin to programming, that is very time-consuming compared to more traditional methods. But maybe I'm just a dumbass.

I like the blobby painted interface in the presentation.

A professional storyboarding artist does a live storyboarding session.

He has a cool art style, and is very good at drawing.

At the end he also mentions the Grease Pencil VR, and says it will be many times faster to storyboard in VR. It’s not yet clear when we’ll have any of this functionality (in Blender, it's already available from other apps).

Another car talk. It’s a pretty promo video that they made for their hypercar. Most of this talk is really an ad for Skoda, but in the second half of the video he goes over some optimization they had to do to be able to use their heavy (3 GB) car model with simulations etc. So they used proxies, baked VDBs, etc...

He claims that they used Blender for 98% and coffee machine for the rest. But it seems they only used Blender for car animation and rendering. 

Not for modeling or compositing. So that’s a pretty amazing coffee machine that can handle those tasks! Perhaps I misunderstood. 

They also used Launch Control add-on for the car animation, which was mentioned in the other car talk as well.

This is a pretty amazing talk. He used Blender, with animation nodes and booleans, to design a huge building with many floors, and then it was built exactly like that. 

Besides for design, they used the 3D model in Blender for virtual walk throughs to show it to the clients. 

He shows some interesting numbers also that you don’t see often. The VFX industry has about 40 billion US dollar of revenue per year. Video games, 220 billion. Construction, 13 trillion US dollars. That's not exactly a little niche sector. 

He talks about some of the tools they use, such as Bonsai, which is apparently a version of Blender specifically for reading and viewing some kind of construction file format (IFC?). https://bonsaibim.org/ 

It’s not exactly clear to me from the talk what their pipeline was, and how and in what capacity Blender was used, but that’s probably because I’m not an architect. I bet this is very interesting for certain architecture students (ones very comfortable with technical stuff like geometry nodes).

glTF is a file format for 3D models. It should be able to contain everything you need: shape keys, rig, materials, etc. I can't honestly say that I've had good experience with it when choosing it for example on Sketchfab. So usually I just go for the FBX file and fix all the problems manually. 

Time to try it again, maybe it works better now. I'll be honest, I didn't watch this entire video. It's a dude talking about a file format, I have my limits 😵. 

This for me is also just one of those "it just has to work, I don't care how" things.

But of course, this is a great initiative. If it really works, it's an almost magical file format. So kudos to them for making it work.

A Blender jam is some people coming together to do some Blendering. Seems like a fun concept! 

Most of this video is just the speaker and audience quietly working on some models. 

So not really interesting to watch (and not really intended as a thing to watch and not participate in), but a fun thing to organize and participate in I'm sure!

This talk is about teaching AI to drive cars. 

It's very similar to the earlier talk about training AI with renders of soda bottles. But with streets and car obstacles. 

Instead of making 100% created artificial scenes, they have a hybrid approach where they add for example some 3D model of a traffic cone or some other obstacle on the road in a live action plate. 

Her advice on the topic of AI at the end of the talk is literally "If you can't beat them, join them". 😬

An interesting talk about the production of a 3D animation called Treasure Hunt, which is animated by friend of Blender Secrets, Pierrick Picaut (see also the video he made for this channel). 

I really like this “chunky” look they developed for the character, with very thick cloth and such, which if I understand correctly was at least partially motivated by the desire to have clean, simple quad topology (no modeled folds and such). 

The tertiary detail is all in displacement maps. The advantages of this are of course: easy to skin, to rig, but also easy to swap the displacement map out for a normal map, so the model can be easily used in a game engine. 

They used storyboards and animatics, and were careful to use a dark environment as contrast to their bright, colourful characters. Geometry Nodes were used for set dressing the environments. 

Interestingly, they found that using linked assets caused a lot of crashing at render time. So they changed instead to one massive blend file with all the stuff in it, which reduced the crashes (if I understood it correctly).

They rendered to Multilayer EXR and used the ACES color space to grade in Resolve.

In the second part of the video, Pierrick gives some animation tips.

Pierrick recommends first blocking out all the shots, which I thought was interesting. This way, you have better consistency of your animation quality, amongst other things.

Very interesting talk if you're going to do some animation or work on an animated short.

Another portfolio presentation of an architecture company who use Blender for visualisation, and also for some designs.