Perhaps... Perhaps...

Awesome to have you in class! ANGM4313 is the class to take if you are interested in FX in the future.

Probably. There are different types of Tech Artists in different studios, but they all tend to have strengths in coding and a thorough understanding of computer graphics principles and the CG pipeline. So, if an individual finds themself spending lots of time gaining mastery in those areas, getting into Tech Art will probably be interesting.

Thanks Harris! The answer is simple: learn Houdini. It's the industry standard for film/animation FX - magic, sparks, explosions, fire, water, etc.

Thanks!

Tech Artists are rare gems to find indeed.

Not that I am aware of at the moment. I am only just getting started in playing around with experiments in Langchain and RAG which would certainly be the right tolls for doing something like that.

i use LLMs primarily for code completion via tabnine, but their codestral implementation can now make a passing attempt at vex. it's supposed to know 80+ languages, and vex is indeed one of them, but it doesn't understand the snippet method, really. but it knows more than previous llms. it's not just guessing at code completion.

@@siftycat Are you using tabnine in an IDE like VSCode and just moving code between the wrangle nodes and your IDE?

@@NelsonLim I use vscode with the tabnine extension as my external editor in Houdini. :) anything that needs more than a couple of lines of vex is easier that way.

@@NelsonLim VS Code with the TabNine plugin as my Houdini external editor. :) Unfortunately it can't really understand your entire Houdini network. It doesn't understand contexts unless you specifically tell it in TabNine chat. Even then, I've deliberately asked it a question I knew, it's given me the wrong answer (wrong number of parameters, not understanding how a snippet iterates over elements), I've then given it the *right* answer and it's gone on to correct me again. I don't often ask it questions - it was just an experiment for stream! :) TabNine's best advantage with any model is being able to read the entire project and understand it, so when I use Codestral for the Python Unreal API in PyCharm, it does a pretty good job despite not being trained on the API. It does seem to have some training on the Unreal C++ API, and the Python API is direct access to that. But obviously if you don't have wider code for it to understand, all you have left is the code completion. Previously it did an okay job because of the C-like syntax of Vex, but now it knows at least *some* functions. But yeah. It also doesn't have the power to understand how we could use functions outside the given context and the amount of actual Vex code out there it can use to train on is slim. I don't have a lot of hope for it outside of code completion, to be honest. That's not just Vex or the stuff we use. That's for everything.

Agreed!

Agreed with you. Houdini artists tend to need a stronger grasp of 3D concepts. You do have a great point, that even experienced 3D artists struggle with 3D concepts if the knowledge is not taught well when learning their 1st 3D package. There is a bit of a rush to quickly get to a pretty image and skip over strong fundamentals that tend to come back and bite in the back when we start to do more advance stuff.

😂 I just needed to trim my nails.

Thanks @ocdvfx!

Good tip! It helps to emphasize the AO that is already present in most GI renderers.

I believe NVidia GPUs support OpenCL. At the moment, I tend to stay with NVidia as some GPU-accelerated tools only use Cuda and not OpenCL.

Thanks. Must be the youthful Asian genes. But I’m not that young. 🤫

I still recommend people watch the keynote if they can. Personally, I felt my coverage and thoughts would appeal to someone who wanted to hear what a professional thought about the features besides the official marketing message.

75 minutes shorter plus valuable commentary

@NelsonLim And that's what I was looking for. But you targeted a time conscious viewer when you mentioned the time saving. I think those that can't free up 2h will also struggle to free up 45 minutes. This is why your statement is a bit comical to me. Now that you've suggested we should add 2h for the original, makes the time-saving comment stand completely out of place.

Big studios of 200 and above head count will tend to use a HP/Dell workstations that often cost 10-20 grand each. Mostly for their corporate support. But most artist won’t need anywhere near that budget to do decent work.

You’re welcome!

Thanks!

I would say that there is lots of good stuff on the Houdini website's learning paths. Personally, I think after going through a tutorial, you should try to implement similar techniques in your own project that is slightly different. Because the essence of Houdini is learning to problem solve. Tutorials are good at introducing concepts, but don't tend to build those problem-solving muscles.

Indeed. Houdini doesn't have much of an edge as a traditional modelling tool. But that's also not where it is used. As a procedural modelling tool though, it is outstanding.

If I were learning rendering in Houdini now, I would focus on Karma.

Have you tried turning the viewport over karma perspective or whatever renderer you are using? The textures don't show up well in HoudiniGL mode.

You’re welcome 😊

Thanks for the suggestion. I'll consider it.

That's so true. It may seem hard to keep up and that there is always something new to learn. I noticed over the years, I have kept to a few foundational nodes, focused on the principles/concepts more and I am more and more curating a set of HDAs/setups that I know won't just change with every new release. So whenever I find something new, I get to choose whether I want to add it to my workflow or not. This might be worth a video, because I can emphatize how much stress it can cause to have to "keep up" with the changes.

Depends on what you are doing. But memory and CPU also play a big part to the Houdini experience. At the end of the day, I've not seen DCC application that can handle as much data and calculations as Houdini and still work so fast.

​@@NelsonLimOn my RTX 3060ti with Ryzen 9 5950x the simulations were simple and slow because I only have 16 GB of RAM 😢

@@robsonhenrique4346 Ah! That explains. Although I’ll say I still pleasantly surprised what my 16GB RAM laptop can do when I have to use it to teach and demo. Unfortunately, it isn’t realistic to expect everyone to have 64 or even 32 GB of RAM. Try joining our Effective Discord channel and perhaps there are ways to optimize your scene to still work reasonably well on 16GB. www.nelsonlim.com/discord

Noted!

It depends. But most are going to be HP/Dell PCs with Data center level CPUs like Epics or Xeons, 256 GB of RAM and an nVidia Quadro graphics card.

You're welcome!

Yes, it is pretty addictive! Many people I speak with agree that once they get it, it's hard to look back.

Thanks. It was a point and shoot Sony EV1.

Agreed.

Almost any monitor can be calibrated. But you'll want to make sure that at the very least you get a monitor that can represent 100% sRGB spectrum.

Early days of youtubing...

I appreciate that!

Houdini technically does not scale the model up. A unit is a unit regardless of 3d application, but 1 unit is Houdini is 1 meter, while 1 unit in Maya is 1 cm by default. Hence the difference.

Thanks Eric!

If there isn't anything you want to keep in the changes of the node (like in the case of the Edit SOP), deleting and putting a new one is probably easiest.

@@NelsonLim What's curious is that the UVedit SOP does have a 'Reset All Changes' parameter button. Maybe I'll file an RFE for the same on the standard Edit SOP : )

Thanks Abhijith!

Hey Felix, thanks for your thoughts. I can feel your frustration. I'm sure plenty of people feel the way you do. Hence, I wanted to provide some of my thoughts on this. Indeed, for folks who came from a software that black-boxed many things from the user, suddenly, they do have to learn many of the CG concepts that was taken for granted. While your experiences are valid, there are also plenty of people who find learning Houdini isn't as hard as others make it out to be. Mainly because they didn't have to unlearn/re-learn CG concepts. They don't feel the same frustration. Houdini is a powerful tool that makes artist develop a better grasp of CG. Some see that as a hindrance while others see it as being liberating. Beginners like me didn't know any better years ago when I started and dove right in and ate it up. I've trained plenty of beginner CG artists in Houdini and I've noticed the same pattern. Their beginner's mindset just allowed them to soak and learn and get good at it quickly. I agree with you that it requires the artist to use both sides of the brain. That is what is truly exciting about Houdini. But no software is for everybody and we do need an eco-system of CG software to help us do our best work. That said, I've also found that the mindset that Houdini creates is one of problem-solving. Something that studios working on the cutting edge of CG look for. Years ago, before Pixar even used Houdini, they started asking for Houdini knowledge in their job description because they knew that the type of problem-solving artists they needed could be found amongst Houdini users. When I was at ILM, the software they used encouraged artists to solve hard and interesting problems to achieve novel looks and visuals in shots that have never been done before. There wasn't always a plugin that would do it for them. Hence, traditional VFX artists are really tinkerers and more like Da Vinci - using their left and right brains simultaneously. I'm glad you are giving it a go because you see the advantages of it. But you do have to approach it with a beginner's mindset. Kind of like learning a different form of martial arts. I don't assume that because I have a taekwondo black belt that I would get my jiu jitsu black belt easily/quickly. It's uncomfortable at first to someone schooled in a different system, but truly liberating when you get good at it and figure out how to use each system to their strengths. I wish you all the best on your CG journey!

Thanks. Unfortunately, I don't use redshift. But you can use the shop_materialpath attribute to assign different materials.

Always welcome

Welcome.

Thanks!

We use both. Just whichever is most convenient. Anecdotally, I generally see more VOP networks then VEX code when looking at Houdini scenes, mainly because noise functions are more intuitive via VOPs.

Yes, Houdini isn't the only tool around. Better to know what you want to create or jobs you want to apply for and learn the right tool(s) for it.

Glad you like them!

Thank you! Cheers!