VLM diary - from a clown to a respected comedian.

[takise]

I decided to start this thread for everyone who is starting their adventure with VPX, blender and VLM in general from scratch.

 

I will keep it in the form of a diary so that in the future new people will have an easier path.

 

In general, I have many years of professional experience in designing for advertising and printing.
I have expert knowledge of Adobe Suite, Corel and several other programs.
When it comes to 3D design, I dealt with Maya, 3D Studio, and Auto CAD.

 

I'm a fan of efficiency and standardization.
So in every situation, my instinct for simplification and automation kicks in.

I will try to find new and easier solutions to existing problems and I will share my observations and thoughts with you.

 

English is my third language and I haven't used it since high school for 20 years. I started using it in speech 8 years ago after I settled in Scandinavia.

So please forgive me when I sometimes say something incomprehensible and please correct me immediately.

 

As the thread develops, I invite anyone interested to talk.

 

Chapter One - Like a virgin.

 

1. Problem nr.1  - IMPORTING TABLE to VLM

 

After watching the tutorial: 
 

 

Installed new blender and VLM and I have a problem with importing the table.

 

After importing, it looks like the attached photo.
The imported table is simply a new project in vpx saved without any modifications.(v2b.png)

Existing finished table loks like (b2b2.png)

 

Has anyone encountered a similar problem and managed to solve it?

 

 

 

Edited January 1 by takise

[michel757]

13 hours ago, takise said:

Installed new blender and VLM and I have a problem with importing the table.

 

After importing, it looks like the attached photo.
The imported table is simply a new project in vpx saved without any modifications.(v2b.png)

Existing finished table loks like (b2b2.png)

 

Has anyone encountered a similar problem and managed to solve it?

 

 

 

Hello,

 

I had the same problem as you in version 4 of blender. I reinstalled version 3.6 and no more problems.

Good year.

Michel757

 

[takise]

Thank You mate i will check this.

 

Happy new year.

[takise]

1. Problem nr.1  - IMPORTING TABLE to VLM

 

Solution:

The solution to the problem was michel757 solution, i.e. reinstalling a lower version of blender, version 3.6.

Thank you for Your help.

 

A moment later, more problems appeared, which I will tell you about soon...

 

After installation, I started watching another tutorial.

 

 

 

First of all, I had to set up the graphics card during the Engine Cycles setup in Blender.

 

First of all, it didn't work - the option in Cycles was INACTIVE.

 

So, as instructed, I opened the EDIT/PREFERENCES menu and selected SYSTEM from the left panel.

And then in the CUDA and OPTIX tabs I selected both items, i.e. both the graphics card and the processor.
 

 

After this operation, the option in the RENDER/Cycles panel was selectable, so I selected GPU Compute (photo1 panel).

 

Of course, along the way, a question arose whether I should select the GPU and CPU in both OPTIX and CUDA?
Or just select one tab?

 

They say that OPTIX is much faster on RTX cards and that it ensures the use of RTX cores for ray tracing calculations.

However, during the first rendering, which I will talk about in a moment, I had both options, CPU and GPU, selected in both CUDA and OPTIX settings.

 

I decided to satisfy my curiosity about efficiency and speed.
I imported JP Salas's Cheetah table into VLM and decided to run it through quick rendering to check the operation of the blender and the plugin...

 

to be continued...

 

 

 

Edited January 1 by takise

[dcbenji]

Select only your GPU in the Optix tab. You can test the difference between Optix and CUDA, but having

CPU also checked in either tab slows things down.

 

Edited January 1 by dcbenji

[takise]

55 minutes ago, dcbenji said:

Select only your GPU in the Optix tab. You can test the difference between Optix and CUDA, but having

CPU also checked in either tab slows things down.

 

 

Hello, thanks for joining and a happy new year.

 

I noticed that when the GPU is marked in optix its automatically unchecked in Cuda.

From what I have noticed, you need to select Denoiser as Optix in the settings of this panel on the right side.

 

I get into the subject, soon new observations.

[takise]

2. Problem nr.2  - EXPORTING TABLE to VPX

Solution: unknown atm.

 

As I mentioned before...

 

I decided to satisfy my curiosity about efficiency and speed.
I imported JP Salas's Cheetah table into VLM and decided to run it through quick rendering to check the operation of the blender and the plugin...

 

It's a simple table, and by the way, one of my favorites. There aren't many elements here, so I decided it would be a good example for the test.

 

I only set the camera, here you need to pay attention to set the camera as active, so select the camera with the left mouse button and then right-click and select Set Active Camera, or use the shortcut cntrl + num 0.

 

My earlier dilemma with choosing CUDA and OPTIX resolved itself.

I noticed that when we mark the GPU in the OPTIX tab, it will automatically be unchecked in the CUDA tab.

If we have an RTX card, Nvidia recommends using OPTIX, so I won't discuss it for now.

The selected option for Cycles or any other engine we decide to use will be OPTIX when we mark GPU usage in its settings.

However, this involves using another setting, OPTIX DENOISER.

 

Right panel, VIEW LAYER/Pases/Data tab, I selected Denoising Data.

 

 

Right panel, RENDER/Sampling/Render tab, I selected Denoise and in the Denoiser position I selected OPTIX from the list.

 

 

Nvidia recommends this in this OPTIX tutorial, it affects denoising in Viewport and significantly affects the quality of the final render...or so they say.

 

 

Ok, so now I've set some rendering values in VLM.

Render height and texture size, both at 2048.

 

 

I opened the console and ran it via Batch All.

The operation took about 80 minutes and... ended in failure.

 

There were no errors during the process, but it did not export the file to VPX, i.e. the table with the VLM note.
I received the final message in the console as shown in the photo below.

 

 

to be continued...

 

 

 

 

Edited January 2 by takise

[michel757]

12 hours ago, takise said:

2. Problem nr.2  - EXPORTING TABLE to VPX

Solution: unknown atm.

 

As I mentioned before...

 

I decided to satisfy my curiosity about efficiency and speed.
I imported JP Salas's Cheetah table into VLM and decided to run it through quick rendering to check the operation of the blender and the plugin...

 

It's a simple table, and by the way, one of my favorites. There aren't many elements here, so I decided it would be a good example for the test.

 

I only set the camera, here you need to pay attention to set the camera as active, so select the camera with the left mouse button and then right-click and select Set Active Camera, or use the shortcut cntrl + num 0.

 

My earlier dilemma with choosing CUDA and OPTIX resolved itself.

I noticed that when we mark the GPU in the OPTIX tab, it will automatically be unchecked in the CUDA tab.

If we have an RTX card, Nvidia recommends using OPTIX, so I won't discuss it for now.

The selected option for Cycles or any other engine we decide to use will be OPTIX when we mark GPU usage in its settings.

However, this involves using another setting, OPTIX DENOISER.

 

Right panel, VIEW LAYER/Pases/Data tab, I selected Denoising Data.

 

 

Right panel, RENDER/Sampling/Render tab, I selected Denoise and in the Denoiser position I selected OPTIX from the list.

 

 

Nvidia recommends this in this OPTIX tutorial, it affects denoising in Viewport and significantly affects the quality of the final render...or so they say.

 

 

Ok, so now I've set some rendering values in VLM.

Render height and texture size, both at 2048.

 

 

I opened the console and ran it via Batch All.

The operation took about 80 minutes and... ended in failure.

 

There were no errors during the process, but it did not export the file to VPX, i.e. the table with the VLM note.
I received the final message in the console as shown in the photo below.

 

 

to be continued...

 

 

 

 

 

Hi,

 

The problem comes from version 0.0.8 of VLM ?with version 0.0.7 the vpx export works.

On the other hand, I had to uninstall then reinstall blender to be able to install version 0.0.7

I will follow your discussion thread with interest, because I am new to VLM

[takise]

1 hour ago, michel757 said:

 

Hi,

 

The problem comes from version 0.0.8 of VLM ?with version 0.0.7 the vpx export works.

On the other hand, I had to uninstall then reinstall blender to be able to install version 0.0.7

I will follow your discussion thread with interest, because I am new to VLM

 

Exactly as you say Michel, i installed 0.8.
I'm glad you're joining the discussion and thank you again for your help.

I'm surprised that there are so few discussions devoted to the technical issues of creating tables.
Generally, there is only a handful of information on this subject and you have to look hard to find anything.

[michel757]

19 minutes ago, takise said:

 

Exactly as you say Michel, i installed 0.8.
I'm glad you're joining the discussion and thank you again for your help.

I'm surprised that there are so few discussions devoted to the technical issues of creating tables.
Generally, there is only a handful of information on this subject and you have to look hard to find anything.

 

Your thinking is very correct.

It is very difficult to find technical information on creating tables.

 

Until now, I was content to adapt existing tables in VR. But I would like to create my first table.

[takise]

18 hours ago, dcbenji said:

Select only your GPU in the Optix tab. You can test the difference between Optix and CUDA, but having

CPU also checked in either tab slows things down.

 

Edited January 2 by takise

[takise]

From over 20 years of experience with various software that I had to learn, I conclude that after half a year we will have more knowledge than what is available on the Internet. And after a year we will be able to write a short book on this topic

 

You have to survive this period during which your mind becomes accustomed and then you start to learn how to use the possibilities.
Blender throws you off from the very first moment with an overwhelming interface.
But this will pass after some time and you will start looking for new options when the blender runs out of them.

 

I installed VLM 0.7, started rendering and waiting for it to finish.
This time I marked only GPU in optix to compare the time with the previous rendering with both options selected.

I will post an update when it finishes baking.

[takise]

2. Problem nr.2  - EXPORTING TABLE to VPX

Solution:

Once again michel757 found a solution to the problem. You should use the older version VLM 0.7 instead of the newer VLM 0.8.

Thank you for Your help again mate.

 

So in short, Blender v3.7 and VLM 0.7.
The table has been baked with Cheetah v0.4 from JP.
I did not make any modifications to the table, it was only a speed and efficiency test.

 

Previously, the same table in VLM 0.8 took almost 90 minutes to bake.
In the blender system settings, I then selected GPU and CPU in OPTIX.

 

Currently, the same table in VLM 0.7 took 58 minutes to bake.

I don't know whether the increase in performance was related to the use of only the GPU option in OPTIX, or whether the newer version of VLM simply performs more calculations due to the additional functions added in it.

I bet on the latter, I think the process in the newer version is more complex... but that's just my guess.

I also used denoising for OPTIX.
I have added screenshots with blender settings at the end of this post.

 

The table size after the VLM process changed from 7MB to up to 56MB.

 

My conclusions and unknowns:

 

- The table is not assembled, there are no ramps or moving animated elements, the texture size and render height are set to 2048, which is low.

Rendering time almost an hour.

I haven't calculated it exactly, but I think that with a complex project in 4K, on RTX4090 and I9 13900K with 64GB Ram, the duration will range from 8-12 hours. In VLM 0.7 in VLM 0.8 this can be 30% longer i guess.

And now I'm wondering whether it is possible to selectively render the corrected elements later?
Or do I have to render everything again to make corrections?

I would like to know whether those who have high-quality tables created using VLM render on their computers or use render farms?

 

The next step will be to make 3 tables, each with a higher level of difficulty.

I mean mod, of course, to obtain the best possible visual effect using available methods.

 

This is to learn the tools, terminology and direct correlation between Blender and VPX.
Finding simplifications and systematizing the work process.

 

I chose three tables that range in increasing complexity accordingly, and I love them.
1. Gottlieb Cheetah
2.Gottlieb Arena
3.Gottlieb Surfn Safari

 

I will tweak each element on each table to see the effects of each change.
And by trial and error, to get an idea of what you should and shouldn't do.

 

I'll start with light because I have no idea about the interplay between flashers, tubes, inserts, etc.
I also don't know what should be done only in VPX and what should be done only in blender.
I'll leave modeling and graphic processing for dessert because I don't have to learn it.

 

I'll start my search with Bord's tutorial on creating Sea Witch and the Flupper tutorial that came with Whirlwind in pdf form.
By the way, this is the best tutorial I've come across so far and I praise him for sharing his knowledge in such a clear form.

 

I will make videos in the form of tutorials every step of the process I will have to deal with... of course not while learning, but when I am sure how to approach a specific problem.

 

Everything will be reflected in this diary.

 

Chapter Two - I start touching him.

 

to be continued...

 

 

 

Edited January 2 by takise

[Thalamus]

It is known that blender 4.0 breaks the current latest version of lightmapper, but, 0.8 should be improved over 0.7, but, I believe you need to download the latest from "actions". He simply hasn't created a release yet for those newest changes. Bare in mind, the lightmapper is very much a WIP, so, expect gremlins. I guess, if you create a simple project, then 0.7 will be fine too. Might be a good idea to look at those videos and make some notes. The naming of objects can be quite important and if you don't have experience with VPX already. Just renaming an object and still make the table work is a task to learn all on its own.

Edited January 2 by Thalamus

[takise]

40 minutes ago, Thalamus said:

It is known that blender 4.0 breaks the current latest version of lightmapper, but, 0.8 should be improved over 0.7, but, I believe you need to download the latest from "actions". He simply hasn't created a release yet for those newest changes. Bare in mind, the lightmapper is very much a WIP, so, expect gremlins. I guess, if you create a simple project, then 0.7 will be fine too. Might be a good idea to look at those videos and make some notes. The naming of objects can be quite important and if you don't have experience with VPX already. Just renaming an object and still make the table work is a task to learn all on its own.

 

Thank You for joining the discussion and your valuable suggestions.

Edited January 2 by takise

[takise]

I'm starting to have fun.
I take apart the table in VPX and start to understand the relationships and role of the elements.

 

Overall, it's an extremely simple and clear editor... I'm surprised because I thought it can be a total mess.

 

The number of questions is growing, but at the same time I am finding answers to the previous ones.
Tomorrow I will try to post a video of the disassembly and analysis of the table elements, which will be an introduction to the editor.

[bord]

Looks like you're making good progress. The lightmapper is still beta and largely undocumented so lots of things to work through, but the results are great. I've done a handful of VLM projects so far and I enjoy it more than the previous process of rendering textures and transferring them to VPX manually.

[takise]

6 minutes ago, bord said:

Looks like you're making good progress. The lightmapper is still beta and largely undocumented so lots of things to work through, but the results are great. I've done a handful of VLM projects so far and I enjoy it more than the previous process of rendering textures and transferring them to VPX manually.

 

Thank you very much for Your visit and kind words.

 

You know, man, you're a king to me, f Elvis of VPX.
I play your tables every day.

I feel like I got an autograph from the King

 

Really big respect bro.

[bord]

2 hours ago, takise said:

 

Thank you very much for Your visit and kind words.

 

You know, man, you're a king to me, f Elvis of VPX.
I play your tables every day.

I feel like I got an autograph from the King

 

Really big respect bro.

 

Very kind of you. But really look at what you're doing. You're powering through a difficult and underdocumented process in order to make awesome games. It is easy to just keep doing things the same way but you're intent on moving things forward. Big respect to that.

[takise]

I found iaakis tutorials on YT. 
I wonder how difficult it is to access these materials online.
When you enter a phrase like "vpx light mapper tutorial" or something like this, not much appears in the search results.
 

I took a quick look at one of them.
It's from a year ago and uses an old version of VLM, but overall it contains a lot of information and is worth watching.

 

 

After watching it, I realized that I needed to arrange a microphone to freely comment on my introduction process.
I'll check if the one I have in my iPad can be used in OBS.

 

The next step, as I mentioned earlier, will be a video of my exploration of the VPX editor.

I'm starting from scratch, so everything is done by asking questions to myself and then constantly trying to find the answer.
The process is easier when you can comment on it in real time.
But I can't imagine describing all this in text form.

 

A new observation and another question, iaaki in his tutorial show how to turn on Cycle in blender (1:23).
He recomended CUDA and select the CPU and GPU in the system.

 

This issue must remain clearly specified.

 

Knowing these types of issues when you work in a 3D program is necessary because of efficiency and speed.

I've worked in 3D programs, but never in Blender... I have no idea about its settings and what changed in blender over the last year.

 

Finally, a technical issue that occurred to me.

 

If there are, for example, 10 light sources on the table that work simultaneously.
All these lights work in two states: on and off.
This means that if they are turned on, they are all turned on at the same time, and they turn off in the same way.
First of all, when we have a table in VPX that we will then import into VLM, all these lights are imported as 10 lights in blender.
We later have them on the VLM Lights layer (or something like that) as layers called gi1-gi10.
What this means during baking is a separate scenario for each of these lights, multiplied by additional interactions with objects.

 

Is it possible to use a solution in a blender that reduces these 10 lights to one light that emits in 10 places?
I also don't know whether all the lights have to be added on the table in the VPX editor or maybe they can be added directly in blender while maintaining the nomenclature used in the VPX script?

I also wonder if objects are used as light emitters and what about light linking?

 

I'm starting to get more and more involved and I'd love to sit on it in the privacy of my home, but unfortunately I have to work... and I'm at work right now

 

... so see you soon.

 

 

[takise]

14 hours ago, bord said:

 

Very kind of you. But really look at what you're doing. You're powering through a difficult and underdocumented process in order to make awesome games. It is easy to just keep doing things the same way but you're intent on moving things forward. Big respect to that.

 

Thank You .

 

I am really interested in the topic, I have been following the development of VPX since its inception, I have always been addicted to pinball
However, I waited patiently because I put emphasis on the visual side of my work.

 

I also think there would be a lot more people interested in starting table design.
However, they are turned off by the barrier to entry into the topic due to the small amount of material on the subject that would introduce them step by step.
Development will accelerate with each new person and new perspective, so the priority should be to encourage as many new people as possible.

So if I finish this topic, maybe I will attract more people like me.

 

The biggest stimulus that made me want to delve into it was Flupper's Dinner, then I realized that there are already opportunities to create projects at a high visual level.

Later, when he published his other tables, I was sure about it.
However, at the time, the amount of work it required put me off.
Additionally, the previous equipment did not cope well with 3D tasks.
Currently, the whole process is much more pleasant, with new GPUs and CPUs you can work on this type of projects at home quite comfortably.

 

I'm not a programmer and that largely limits my point of view on many issues.
I have to approach everything from a designer's perspective and visualize most problems.

 

However, as they say, a real man is determined by how he finishes, not by how he starts... so time will tell.

[bord]

1 hour ago, takise said:

Finally, a technical issue that occurred to me.

 

If there are, for example, 10 light sources on the table that work simultaneously.
All these lights work in two states: on and off.
This means that if they are turned on, they are all turned on at the same time, and they turn off in the same way.
First of all, when we have a table in VPX that we will then import into VLM, all these lights are imported as 10 lights in blender.
We later have them on the VLM Lights layer (or something like that) as layers called gi1-gi10.
What this means during baking is a separate scenario for each of these lights, multiplied by additional interactions with objects.

 

Is it possible to use a solution in a blender that reduces these 10 lights to one light that emits in 10 places?
I also don't know whether all the lights have to be added on the table in the VPX editor or maybe they can be added directly in blender while maintaining the nomenclature used in the VPX script?

I also wonder if objects are used as light emitters and what about light linking?

 

 

You are understanding this correctly. If the GI lights in Blender are each assigned a corresponding name they will render separately, increasing render time and system load. You can change the name of each Blender light to share the name of a single VPX light and they will render together. BUT the cost is that the raytraced ball shadows in VPX 10.8 will not work properly. I tend to bulk render the GI lights where this might not matter but render separately the ones where they ball travels nearby often. The visual effect is worth it.

[takise]

1 hour ago, bord said:

 

You are understanding this correctly. If the GI lights in Blender are each assigned a corresponding name they will render separately, increasing render time and system load. You can change the name of each Blender light to share the name of a single VPX light and they will render together. BUT the cost is that the raytraced ball shadows in VPX 10.8 will not work properly. I tend to bulk render the GI lights where this might not matter but render separately the ones where they ball travels nearby often. The visual effect is worth it.

 

I understand Your answer, but I don't understand why this is happening in terms of VPX operation.

 

If we generate this group of lights as one light and the ball is within the range of just one of these individual light sources, will  raytraced ball shadows be generated correctly?

Does distortion only occur when interacting simultaneously with more than one of these component sources of our single light at the same time?
What I mean is that, for example, these three small light components are close to each other and the ball enters their simultaneous range.

 

I have doubts whether my English is understandable enough.

 

You don't have to answer, I'm not going to over-absorb you, I'm just thinking out loud.

Thank You for your earlier reply.

[bord]

3 hours ago, takise said:

 

I understand Your answer, but I don't understand why this is happening in terms of VPX operation.

 

If we generate this group of lights as one light and the ball is within the range of just one of these individual light sources, will  raytraced ball shadows be generated correctly?

Does distortion only occur when interacting simultaneously with more than one of these component sources of our single light at the same time?
What I mean is that, for example, these three small light components are close to each other and the ball enters their simultaneous range.

 

I have doubts whether my English is understandable enough.

 

You don't have to answer, I'm not going to over-absorb you, I'm just thinking out loud.

Thank You for your earlier reply.

 

Your English is quite good. If the bulbs are very close together it might not be noticeable. In the image below I tried to illustrate the two options by tracing out how VPX would map out a shadow based on ball position.

 

[takise]

6 minutes ago, bord said:

 

Your English is quite good. If the bulbs are very close together it might not be noticeable. In the image below I tried to illustrate the two options by tracing out how VPX would map out a shadow based on ball position.

 

 

Once again, thank you for Your answer.

 

With your beautiful answer, you visualized exactly what I had in mind.

 

Thank you again for your kindness.
I promise not to bother you again