Contribute!

 

Silver!

This morning, I'm fighting pain and lack of sleep. It's down to a partly completed root canal surgery, which has left me with the prospect of either a tooth extraction next week, or some kind of special hand-wave procedure that sounds like Scotty from Star Trek thinking up some technobabble solution to avoid antimatter annihilation. 

In this weakened and emotional state, I completed the 2025 Blender Survey, which prompted me to reflect on my lack of contribution over the past few years. I stopped my subscription years ago, as part of an urgent cost-saving measure. I later excused not restarting payments because I wasn't doing as much Blender work. Later, when Blender became my "main thing", I felt that I was spending a lot of money on Blender Addons, which adds a small contribution with each purchase. I decided that I could afford £10 per month in addition to all the SuperHive store purchases. That makes me feel better. Maybe this excruciating pain was really the build-up of guilt for not contributing? *Side of my face still throbs with pain* Nope, it really is a swollen gum. 

Go on, give them some money!

Release Candidate

Blender 5.0 Release Candidate snuck out in the dead of night. I installed it on my secondary machine, but I will hold off installing it on my main machine. A major release means a fresh installation and the case-by-case reinstallation of addons. It's potentially troublesome, and while I thought I'd be able to do it over this "candidate week", some addons will still need updates and probably won't appear until after 5.0 enters general availability, next week. I will do some play testing, but I'm not going to risk falling into a can of horrible software worms in my current state.

Well, hello there!

Well, Character Creator 4 still works, though it's installed on my secondary machine.

Cthulhu cubes -- generative AI assets.

Egg house progress. I have 1 week until the celebration scene needs to be finished.

Will we see a new candidate?

 It's the proposed date of the Blender 5.0 release candidate. Will we see it today? I've noticed that another patch is due for the 4.5 LTS version, with a release candidate for 4.5.5 already in testing. Does that impact the 5.0 release? We will know later today.

Yesterday's tooth pain had to running to my therapy space: terrain sculpting and rendering.

Dentist = Ouch

 This morning's dental surgery has left me in some considerable pain. My ability to output anything worthwhile is being curtailed. 

Luckily I did some more cool things on Monday.


Egg House progress

The weird egg-happed lifepod-turned-emergency-shelter thing is progressing. I started to play with materials as I thought about what else it needs.

All the materials are from the Sanctus Library, which remains a brilliant resource, but you may need to go deeper for true "hero" materials. The outer shell should be really realistic, with additional layers of grime and some graffiti. To be developed further! I should get the interior modelling done first. 

I also want the egg to be dressed with cables and rough, crafted elements, such as an awning and a washing line. This pod appears to have floated down from the sky and been stuck here for years, gathering moss and becoming the centre of a primitive habitation.

Typical weekend

 Was it a lazy weekend? Well, I didn't do very much. Battlefield 6 was offering double-XP, so it made sense to mindlessly waste all my time on that. But not all my time, really. I actually did some cool things.

I did a lot of scatter practice, as I want to do a really good job on the Blender 5.0 celebration render.

New terrain. Lots and lots of new terrains

Tres dark out there.  It's hard to see the biome test.

I played with GeoScatter's biome system. Biomes are a collection of ground scatter recipes that carefully layer different ground assets to produce an instant environment type. They are too resource-hungry to use on anything other than a tiny area of foreground. 

Biomes in a pot

From now on, I'll use this smallish pot (it's enormous, for a pot, but very small as a terrain) to test complex scatter collections and own-grown biomes.

A dry grass terrain (my own scatter collection)

Arid (Bagapie biome collection)

Customised biome

Customised biome

Another terrain and some of those wonky generative AI assets

This element looks pretty cool. I will generate some complementary elements to form a full Lovecraftian alien temple.

This week

Blender 5.0 Release Candidate

This will set the clock for the full release in the following week, unless something stinky turns up while testing. I will migrate some of my own work to 5.0, although project work, such as the Service Road, must remain on 4.5 LTS.

True Sky 3

Oh boy, this looks absolutely amazing, especially with the added sky physics and support for planets.

Layered clouds and a "projected planetary sphere"

Service Road

Complete the requested changes to the service road elements. Probably a day's work all in all.

Celebration Render

Complete the modelling and texturing of the egg house. Select a terrain and start dressing it. I had trouble testing BagaRiver addon. I might look into it as it looks like a great way to add a nice stream. This might be handled by Nature Generator; I'll consider that, too.

Service Road Updates

 Last night, I had a meeting with Andrew, the director of the animation project. He had a list of alternations that I need to do on the submitted service road scenes. It's mostly fine detail stuff where the scene doesn't quite work with the character staging. There's only one significant change, where the "sniper alley" section of the road needs to be straightened to deny characters any obvious and easy places to hide along the valley walls. It's not so big that it requires a complete rebuild, although I will do a rebuild of the scatter systems so they're consistent between sections.

That will be next week's work. 

Weekend

I will be playing around with the Blender 5.0 celebration scene and probably wasting time on "terrain therapy".

Issues of the worrying kind

I've had two recent incidents where my PC's display went black. This issue was reported early on in the release of the 50-series GPU card. I thought it was fixed, but it may be back, unless this is something different.

Next week

Blender 5.0 Release Candidate

I'll start testing it, although project work may continue in 4.5.3 for a white. Full release will be the week after. 

True Sky 3.0 should be released. It looks like an amazing upgrade. Space, planets, rings, multiple cloud layers. So excited!

Supposed to be grassy. Turned out Lovecraftian

Release day

I placed a hex on today's release by talking about Blender 5.0 in yesterday's post. The foundation just pushed the release back a week, so the release candidate is now due on the 12th and the full release is now due on the 18th of November. Oh hum! These things happen. I suppose the remaining bugs and issues appear too serious to be resolved with a few quick updates over a few days.

I'm still working towards a release celebration render.

Work-in-progress:

Test shot

A render to celebrate the release of Blender 5.0

 A major release (x.x to y.0) for Blender is a big deal!

I've been using Blender for 13 years, and I've only seen two major releases: 3.0 and 4.0.

Blender 3.0 (December 3, 2021)

Blender 3.0 key features (that impact me):

• Cycles X
• Asset Browser
• First geometry nodes

This was a real game-changer, but I was spending too much time gaming, so it didn't impact me very much. In fact, I didn't remember that splash screen. I do remember that Cycles X was a huge improvement. If you are going to make an improvement to a 3d application, you can't do better than make the render engine all-around better. I think that 3.0 would have been the most consequential release in Blender's history, except that late 2.x releases included changes that should have been reserved for a major release. Ton had made commitments about what the next major release would include and felt that moving to 3.0 without these promised features would have felt like cheating the community.

Blender 4.0 (October 16, 2023)

Blender 4.0 key features

• Light and Shadow linking
• Principled BSDF and AGF colour transform
• Geometry nodes improvements

Blender 4.0 felt incremental. Lots of minor improvements without a major new feature, although light and shadow linking had been a big ask from the community for a very long time because it's a standard feature in many other applications.

These summaries overlook some incredible developments in EEVEE, Grease Pencil, and the Compositor, because I haven't used them much myself.

Blender 5.0

Blender 5.0 is available in Beta, but for someone who uses a bucket-load of addons, Beta releases are not much fun. It could be, in fact, that I choose not to move over to the new release for a while, as my critical addons become compatible. True Sky is one such addon, and it's also getting a big update very soon, although apparently this isn't tied to the Blender 5.0 release.

Celebration

It doesn't seem right to create a Blender 5.0 celebration render while not using that version. But it is what it is. 

Idea: A space craft landing pod that has crashed into a forest and has been converted into a habitat. Portrait format, for a change.

Midjourney concept image 1

Midjourney concept image 2

Back to the old grindstone

This morning I started with some housekeeping. I updated to Blender 4.x's final update (4.5.4 LTS) before the provisional release of 5.0, in two days.

Rendering in tiles

I looked into using non-default tile sizes for rendering. It seems intuitive to use smaller tiles to use less VRAM when rendering. I'm not sure if this makes sense when running an RTX 5090 with its 32GB of VRAM. Currently, the use of VRAM during rendering is obscured by a limitation of the Vulkan API, which can't track usage.

Tile size: 1024x1024 

Tile size: 2048x2048

In a very quick "off-the-cuff" test of my current scene, we see that a tiny amount of render time and memory was saved by halving the tile size.

Let's try using Persistent Data, where Blender stores render calculations for reuse in follow-up renders. This might be good if, like me, you do an awful lot of iterative renders.

First render with Persistent Data

The follow-on render took seven minutes and five seconds. A catastrophic impact on render time, probably indicating that no clearing out render calculations exhausted available memory, resulting in a drop in performance. For a smaller scene, you can save part of the rendering process and save some time. Not here, though.

Tiles: 512x512

Tiles: 2048x2048 (Default)

So the smaller tiles saved less than ten seconds. That could be important for an animation where a few seconds will stack up over the total render time, but for stills, not huge. I'll stay with the default until I encounter a problem, then try rendering with a smaller tile size.

Back from the Kingdom of the Netherlands

 We visited my sister and her family because they won't be making their usual trip back to see us during the Christmas holidays. It was a smashing break!

Amster-home-of-Blender-Dam!

Had I not come with the family, I'd have made the pilgrimage to North Amsterdam to visit the entrance of the Blender Foundation. Only a thirty-minute walk from the ferry. I'll do it on a future visit.

Leo and Clara, paddling in the North Sea.

Sand-a-blowing

Den Helder lighthouse

What next?

Complete the pre-vis elements of the service road. In the next day or two, I'll go through some required tweaks with the director.

5's almost Alive!

Wednesday sees the release of the Blender 5.0 release candidate. I'll start using it then. I'll create a piece of work to celebrate the release... More on that tomorrow.

Some more terrain stuff.

 

One from the tres noir genre.

Specularity gives you a water effect.

I've started pushing the size of my terrain files up to the 1GB (8k-by-8k) single-tile export limit. You don't get a lot from the 2x2 increase in size. Although times four sounds like a big increase, you don't end up with a dramatic increase in surface detailing. Maybe a tiny bit. 

Blender School: Maarten Nauta's landscape lessons

 This Blender School lesson will be dedicated to some of the pearls of wisdom that brilliant landscape artist, Maarten Nauta scatters during his Youtube videos (example: Creating Realistic Environments with Blender.)

Preproduction and reference

Golden rule: DO NOT JUST RIGHT IN.
Write down what you need from an effort. Creating an animation will require a different starting point and a different outcome. Set boundries for the client, even if you are the client. Dawn shot? Weather? Biomes? Where is the camera? Where will the camera travel? The more things you nail down, the less chance that your effort goes astray.

1. Project Scope
2. Focus Points
3. Solid Concept
4. Get Your asset library

Reference

For more authentic biomes, consider using iNaturalist to build a shopping list of authentic plant assets.

inaturalist.org is a great souce of real-world biome data.

Maarten - "One benefit of inaturalist is that it gives you photo references from your chosen biome. This is important because the look of plants can change depending on the biome. For example, A Scots Pine in the Baltics looks different in Mediteranian environment. It's an extra step but it can help.

Reference with a mood board.

PureRef is a great tool for quickly creating a reference board.

PureRef (In this case, for Halo Tech)

Golden Rule: Animate at the beginning

If you are creating an animation, don't wait until your detailed landscape is built before doing the first pass of your animation. it will be faster and more stable if you get it blocked out while your scene is small and simple.

Big terrains and grass

In Maarten's example, his terrains are more then 20km front-to-back. Huge! When you apply a grass shader to a large terrain then roughness never looks right. The shader is treating your grass surface as a flat plane, which it absolutely isn't. Grass scatters light really well, it only reflects in exceptional circumstances, such as when it is very wet.

Turn the specularity down to 0.01 or 0.001. You could almost turn it off.

More from Maarten:
"Basically what I'm trying to say is that roughness only dictates how directly light gets reflected. At a high roughness lots the light scatters in lots of different directions causing the reflective sharpness to lower. However roughness of a texture can't emulate the 'absorbtion' (don't know a better word to describe it) of lightrays that happens with lots of foliage. Turning down the specularity lowers the amount of light that gets reflected which kind of emulates what would happen when a lightray hits a real 3D patch of grass rather than a flat grass texture."

Atmospherics

Volumes are the best. They are expensive, result in noise and flickering, but they just give you the realism. You will need high samples. 500-1000.

Anistropics:  0.5 More cinematic, as it congrigates the density of the volume around the lighter parts of a scene. When you use a Gobo, you get a deeper and complex and potentially realistic scene.

Flickering from noise reduction: Consider rendering as separate passes, so that you can apply tweaks to your volume pass, such as a slight blur.

Trees

Maarten uses Speedtree. It's Unreal-based and deeply propriatory -- as in the developer hasn't made it Blender-friendly. Obviously you can still use it.

Tip: Trees that are going into a forest, consider stripping their mesh of lower branches as they will be hidden within a forest. Leave the full branches for forest edges and isolated trees.
 

Optimise your background assets!

Don't use full detailed models in the viewport. Even after years of improvements to Blender's viewport performance, and the move to Vulkan has made a massive different, it still pays to use simple proxy objects in the viewport.

Vegetation and noise

Rescaling plants and trees for realism is very important. However, using a noise for this is not realistic. Vegetation is not random, the noise it presents are based on rules concerning moisture and the neighbouring plants. 

Getting good vegetation is about applying rules instead of random.

Geo-Scatter

Maarten uses manual scatter more than is recommended because it provides the artist with direct control over where stuff scatters. "Easy and optimised". 
Consider using the Reduced Density, right up to 90%

Scattering grasses

Maarten hides this wizardry behind his Patreon. Apparently the knowledge was gleaned from Max tutorials from industry artists.

One this is for sure YOU CAN'T SCATTER realistically on big terrains. You have to understand how hinting at distance grass is the only possibly method. Many millions of grass scatters on top of everything else will crash your system or take too long to render.

Tip: Consider baking terrains into meshes that you can decimate. Subdivision modifier is a memory hog and can be unstable.


Tip: Render with Tiles to reduce VRAM 

Cliffs and terrain verticality

Any displacement map-based terrain is going to suck and areas of steepness. A displacement can't express detail in these areas. If you need detail then you can 1) do a massive manual refinement of the mess, manually adding more geo' into the baked mesh. 2) Add some non-displaced or differently displaced assets into the cliff area.

Stolen from Maarten's Rise and Fall breakdown

The cliff details look like they perfectly follow the underlying terrain, which makes me thing this element has been modified manually to add the vertical displacement.

Maarten's Rise and Fall breakdown - How good is that? Very damned good!