Jump to content

GDTF Forum

ROBE Lighting GDTF files


Petr Vanek - Robe

Recommended Posts

Dear all,

we have been producing tests files for a long time and have switched into a final version  releases now. What we are aiming for is about one library a week. What takes longest is physical attributes measurements (movement, strobing, iris...) and testing to match visualization. The rest - models preparation, geometries, media content, color-metric data, DMX... is possible to do in about two/three days.

You can find our files under our official account:

https://gdtf-share.com/user.php?name=Robe+Lighting+s.r.o.&page=fixtures

And what we are missing is feedback - we do receive some from our customers, but are open to further information in case anything strange/bad is found, so please feel free to report here or to our dedicated email libraries@robe.cz .

thank you

 

 

Edited by petrvanek
new url
  • Like 1
Link to comment
Share on other sites

We have uploaded updated Robin megaPointe, Robin LEDWash 800X, Robin SilverScan and Robin Viva CMY to the share this week. Also, i have created a short video screen-cast on how to prepare 3D models for usage in the GDTF Builder, by using only freely available tools, see it here: http://spares.robe.cz/static/images/gdtf_video.html

EDIT: Added link to newer video, link now points to a page with embedded player.

Link to comment
Share on other sites

Strange. Do right click → Save link as. I will upload it to YouTube or Vimeo. Anyways, i made a small mistake (didn't rotate the head one more time) so had to do a bit of time travel via copy/paste while editing, so am thinking to re-capture it, that is why i didn't upload it to any of the platforms yet...

  • Like 1
Link to comment
Share on other sites

  • 1 month later...

Hi David,

actually yes, i did, but i didn't manage to do the editing and producing it. I was not sure if this is interesting to someone, because it is quite manufacturer specific flow: have an existing model and "only" do modifications, optimizations to it, specifically to GDTF.

Also, fun fact, depending on exact inside rotation of the 3D model, my "quick rotations" done via swapping axes is, while 99% the same, occasionally 90° off, meaning i must use different axes for the swap/rotate.

Yes, this is Linux, i only reboot to Win when i want to see GDTF in MA3 3D or Vision, so that is a bit clumsy, although at times i use a second computer for just that, because rebooting into strange OSes is not fun. Maybe one day we'll get Linux build of MA3 :))

As for the tools i use, Freecad is really good and usable, it has quirks, but as i do not do complex editing but mostly splitting or extracting parts it is very good. So i can for example take only part of a given 3D file, like some of the front lenses, motion camera or other. The rest is done via Meshlab, where i "only" do the rotations, origin setting and mainly the mesh simplification + export. I have not found how... invested time into making this all work only in one of them or for example in Blender but there is also an issue with importing, export to 3DS and other... the workflow i have has been very fast, stable and predictable (key factors for me).

The mesh simplification is a key for GDTF. It took me some time to find Meshlab which does all these things in a fast way. There is Decimate modifier in Blender, and although i have done some (mostly animation) work in Blender, i didn't try using Blender in my GDTF work flow. I am sure C4D has something similar. Mesh simplification is really important. This is because for the most, many people presume that the model has to be as precise and as "nice" as possible, while the opposite is required for 99% of the time and we have been getting a lot of heat for "your models are too complex" over the years - if you patch few hundreds of single unit with model too detailed, planning/pre-viz tools slow down to crawl... I am sure there are custom projects where one wants to build a real-world render with textures, smooth models... and, GDTF is extensible enough to make it possible to use for that and even more in the future (for example by allowing multiple versions of models, adding textures...) , but the initial focus has been on the most required work-flow, so our [Robe] model optimizations are hitting exactly that.

?

P.

 

  • Like 1
Link to comment
Share on other sites

C4D has the Polygon Reduction tool....great set of Axis Tools as well, when you combine with Snap, FTW!  I've already seen bloated models in the Share sites, the format does specify a max count on 3D objects...which you can adhere to and still maintain a respectable shape/model.  To me, it's more the correct scale of parts in relation to each other that provide the detail, not gobs of polys/verts.  I'm seeing some peddled on the interWeb that demonstrate that very shortcoming; poly bloat for the return in the render/visualizer.

 

 

  • Like 1
Link to comment
Share on other sites

I'm with Rex on everything above. Creating good looking models is all fine and dandy, but as Petr stated as more of the larger poly models get added, it is gonna hit the resources of the machine harder. There's no need to have detail that  is not neccessary to the output of the beam of the profiles, the body parts of the fixture should definitely be proportionate to each other and replicate the fixture the best it can to an extent, but unless you're doing rendering's, once the scene goes dark for programming, the need is not there. We build all of our models from scratch in C4D and have been able to severely lose poly's and vertices without degrading the "profile" of the fixtue body parts. If everything is built correctly, one should be able to handle over 1000 fixtures and a full scenic model build out without too much FPS loss pending hardware.   

  • Like 2
Link to comment
Share on other sites

Worked great now, clean, no audio.  Too bad about the pipeline--OBJ--3DS, 2 programs, reorient...ugh.  C4D is fairly inexpensive for what it can deliver....

When I develop 3D fixtures; I usually have C4D open, an IDE for the .XML scripting open, and MA2 onPC/MA3D open....that's enough[4 open programs] on my desktop!

Cheers!

Rex

  • Like 1
Link to comment
Share on other sites

  • 2 weeks later...
On 4/7/2020 at 12:11 AM, Hanz_Brech said:

Hello i would be intressted in the photometric data provided and included in the gdtf files compared with ies files ?

Hello @Hanz_Brech,

thank you for your interest.

The provided spectral data, provided in GDTF right now, are about intensity, color and color spectrum. IES files are about light/intensity distribution.

What you do with the data and how you apply is completely up to you. So feel free to grab the provided files and cross compare or apply in any other way.

Kind regards

Petr

Link to comment
Share on other sites

  • 2 weeks later...
On 4/8/2020 at 6:36 AM, petrvanek said:

Hello @Hanz_Brech,

thank you for your interest.

The provided spectral data, provided in GDTF right now, are about intensity, color and color spectrum. IES files are about light/intensity distribution.

What you do with the data and how you apply is completely up to you. So feel free to grab the provided files and cross compare or apply in any other way.

Kind regards

Petr 

just a quick followup, is the gdtf data of intensity the "same" like ies light/intensity so if u have gdtf u dont need ies or would it be better to combind both for a optimal solution/ more acrurancy?

Link to comment
Share on other sites

Hi @Hanz_Brech,

the intensity in beam and in corresponding IES should be the same. IES gives you much more precision and information → this being details how the measurements have been taken and also about the field of the beam. But then, it depends on how much precision is needed in your application. More precision can also also more calculating power...

cheers

P.

Link to comment
Share on other sites

  • 4 weeks later...

Hi @petrvanek

I looked at some Robe fixtures to see on what geometry you put different functions.... I got confused.

Robin BMFL Spot has all functions in the Head except Pan

Robin Esprite has control channels in the base - which make sense, but also Animation wheel...

Maybe it doesn't matter, but it's not very consistent and not helping me to find best practice.

What do these choices affect? 

Link to comment
Share on other sites

Hi @Ringen,

at this point, for traditional moving heads, the geometry selection dot not affect much. It is with multi "level", multi pixel fixtures, where this is part of the elementary structure how the file is built and it is important. Some of our files had historically been done differently and they could contain some previously used structure. Thank you for noticing and reporting, when doing some modifications, we will change the mapping.

Cheers

Petr

Link to comment
Share on other sites

  • 2 years later...

All currently available Robe GDTF files have been converted to use the glTF mesh format, in the form of glb files. You can read more about it here https://gdtf-share.com/user-page?name=Robe+Lighting+s.r.o.&page=home . This change has happened already in October/November, together with many other improvements and changes like converting the Connectors into Wiring Objects, adding 2D and also including the above mentioned more streamlined linking of channels to particular geometries. Let us know in case of any issues.

  • Like 1
Link to comment
Share on other sites

28 minutes ago, Petr Vanek said:

All currently available Robe GDTF files have been converted to use the glTF mesh format,

 

28 minutes ago, Petr Vanek said:

Let us know in case of any issues.

The current version of the Robin Spiider is being visualised as primitive types only on the current version (v1.8.8.2) of MA3 onPC (see attached). Also note thatonly the center pixel is being visualised, not being too familiar with GDTF or MA3 it's unclear to me if that is simply the result of the visualised primitive types or a separate issue.

According to the share, the 3D model has been tested, how is this being tested?

I do wonder about the benefits of creating fixture specific models, it seems that most are very low quality significantly lacking definition of the details to such an extent the model is barely (and in some cases, totally) unrecognisable. I understand the main purpose of the default res of the model is for use by visualisers and as such super detailed models are not required, but, the max 1200 vertex restriction is too limiting in many cases.

robe-spiider-ma-3d-primitives.png

Link to comment
Share on other sites

3 hours ago, redmonkey said:

 

The current version of the Robin Spiider is being visualised as primitive types only on the current version (v1.8.8.2) of MA3 onPC (see attached). Also note thatonly the center pixel is being visualised, not being too familiar with GDTF or MA3 it's unclear to me if that is simply the result of the visualised primitive types or a separate issue.

According to the share, the 3D model has been tested, how is this being tested?

I do wonder about the benefits of creating fixture specific models, it seems that most are very low quality significantly lacking definition of the details to such an extent the model is barely (and in some cases, totally) unrecognisable. I understand the main purpose of the default res of the model is for use by visualisers and as such super detailed models are not required, but, the max 1200 vertex restriction is too limiting in many cases.

 

> not being too familiar with GDTF or MA3 it's unclear to me

Thank you for your feedback, much appreciated. The GDTF provided limit is more then enough for us to provide nice models and also the lens shapes. And if we needed more, there are now three levels of details in GDTF itself to provide really high quality meshes. The Spiider  uses less then 900 vertices:

models/gltf/yoke.glb 202
models/gltf/head.glb 302
models/gltf/pixel3.glb 20
models/gltf/base.glb 298
models/gltf/pixel2.glb 10
models/gltf/pixel1.glb 12
total vertex count: 844

As for MA, you need to get at least a bit familiar with how to use the software you want to use for the testing. MA requires you to change level of detail for rendering, as sometimes  they change the models to look like packing boxes. Once the setting is done, what you will see is this:

image.png.4a0520fc54d1285e21c8839f9e48b9ce.png

When at full, they look like this:

image.png.c8392796f87359338e16c419a4888db7.png

Hope this helps,

Petr

 

Link to comment
Share on other sites

3 hours ago, Petr Vanek said:

total vertex count: 844

I don't believe this vertex count to be correct, as I don't believe that reusing model files and/or geometry referencing comes for free, does it? pixel3.glb may contain 20 vertices but it's referenced 12 times in the geometry so I don't believe that it adds just 20 to the vertex count, my testing would suggest that it adds 12 x 20 (240). Similarly, pixel2.glb is referenced 6 times so it adds 6 x 10 (60) to the vertex count and pixel1.glb twice so adds 24 to the vertex count. The actual vertex count from the model files is:

models/gltf/yoke.glb 202
models/gltf/head.glb 302
models/gltf/pixel3.glb 20 x 12 (240)
models/gltf/base.glb 298
models/gltf/pixel2.glb 10 x 6 (60)
models/gltf/pixel1.glb 12 x 2 (24)
total vertex count: 1126

However, this total does not take into account vertices from any additional geometry, the "Pigtail" for example, this also does not come for free with regards the vertex count. I haven't seen anything within the standard that details what the max vertex count should be for any of the internally or auto generated primitives, it seems entirely arbitrary and down to the individual consumers to use as simple or complex objects as they see fit.

3 hours ago, Petr Vanek said:

As for MA, you need to get at least a bit familiar with how to use the software you want to use for the testing. MA requires you to change level of detail for rendering, as sometimes  they change the models to look like packing boxes.

For sure I'm very much at beginner/entry level when it comes to all things MA and I could be wrong, however, I don't believe that is the issue in this case. My testing suggests that the MA-3D window with "out of the box" default render settings will visualise models with a vertex count just over 1200 (if my test count is correct, 1210 seems to be the max), anything over that then the model is visualised as simple primitives ("packing boxes" as you term it). I can reliably trigger the "packing box" look by doing nothing more than increasing the vertex count which leads me to conclude that changing (upping) the level of detail for rendering does not, as far as I can tell, enable visualisation of external model files but rather enables visualisation of models with higher vertex counts.

I took the Spiider GDTF file and removed the Pigtail from the geometry and then the model files are visualised with the default render settings which leads me to conclude that the internally generated pigtail primitive adds a significant amount of vertices to any geometry that includes it and, in the case of the Spiider model, tips the vertex count over the max count defined by the GDTF standard for default resolution of 3D models. Either that or there is some issue with the Pigtail definition that's giving MA-3D a hard time but I didn't see anything untoward from a casual glance. 

See below for a screenshot of MA-3D with default settings, your Spiider model above, my Spiider model below. My Spiider model files accumulate to a total vertex count of 1186. If my conclusion is correct that the "packing box" look is triggered by vertex count then it makes no sense that a model with a total vertex count of 1186 is visualised correctly while model files that you say total 844 vertices are being visualised as simple primitives, if I'm incorrect in my conclusion I would like to know what exactly does trigger the packing box look. Does anyone know if there is there any logfile that MA3 produces that might provide some details on this?

gdtf-models-robe-vs-internal.png

Link to comment
Share on other sites

Thank you for confirming that the Spiider comes up OK for you. At the end of the day, we don't control software defaults and even if we try to micro-optimize the Spiider, if you for example add Tarrantula you will have the same issue, so LOD setting will be needed anyways. As far as details of the multiplixel multilayer based devices, this is work in progress for better visualization. You will have to ask MA about the details of their software.

Petr

Link to comment
Share on other sites

The Spiider only comes up if I allow MA-3D to visualise models with vertex counts >1200, this should not be a requirement to visualise GDTF models as the total vertex count for any one model should not exceed 1200. As such I don't consider it correct to say it comes up "OK".

11 hours ago, redmonkey said:

leads me to conclude that the internally generated pigtail primitive adds a significant amount of vertices to any geometry that includes it

I prodded the system a little more on this and have determined that the pigtail primitive generated by the current version of MA3 onPC consists of 190 vertices. Adding this 190 to the total vertex count of the Spiider models (1126 including geometry references) results in the Spiider having a total vertex count of 1316 which explains why MA-3D (with default settings) visualises the Spiider as a collection of simple primitives.

Link to comment
Share on other sites

Interesting... I cannot see where i can set the limit to be below/above 1200 vertices in MA, i only can see Standard/High/... settings, including in their documentation, where can you see this? As per your and mine calculations, the total vertex count of what we provide for the Spiider is still below the recommended 1200 limit, considering the pigtail should be a generic cube, which is what we add when we create the file. On the side note, the GDTF recommended 1200 limit is for a traditional moving head with base, yoke, head and one beam, rather then for a fixture with 19 beams. It is a good conversation, but as this is about specific software behavior, i would suggest that you discuss this behavior of the MA software with MA as i have no inner knowledge of their system...

Link to comment
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.