Sometimes I feel that IRAY or DAZ is just intended for portrait only renders
Cyberdene
Posts: 72
I often wonder if IRAY will ever allow us to populate these scenes with actual figures someday...I can't help but feel that IRay is intended for portrait only renders. The limitations makes it rather imposible for me to do the type of renders I'd like to do. Billboards is a not a good altertantive either. Its like whenever Stone Mason makes a large environment, I have to stop and wonder what is the point in buying them when you can't even populate them with enough people. Every scene I want to capture I get this "End of the world" feeling because these scenes can't even hold the amount of figures in them they should....
This is why I've been hesistate with buying large environments, and do feel that their pointless and a waste since you can't take full advantage of them, and no one in DAZ or Nvidia seems to be working on a way to fix this limitation issue either. You would have to abuse the crap out of layering in Photoshop just to populate these scenes by rendering one figure at a time, then blending them into the scene. It's tedious. I had the same problem with a bar scene. It's like "Damn, why cant I do a party scene with a bunch of people having fun?" You literately would need a graphic card with over 4000GB just to do the type of insane things I'd be doing if it wasnt for these irritating disappointing limitations.
I find no enjoyment what so ever out of doing fancy portrait renders when my interest has always been big story driven scenes with a lot of action and activity going on. If video games can handle that level of detail, there is no excuse what so ever for why IRAY cannot do this. Especially since those are moving images, and we're just rendering stuff. But maybe studios actually have access to superior software, wish there were better alternatives for this. Unfortunately I suppose there is not much we can do to make IRAY more fun...
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2026 Daz Productions Inc. All Rights Reserved.
Comments
I am curious about how Daz3D and Iray handles memory, really annoying as there does not seem to be a reliable measurment within the Daz software. If I have 10 characters using the same textures, does Iray need 10x the texture memory storage of one character? It seems to be the case and it's nuts.
It's very challenging to write software that is always checking when something is supposed to be loaded to see if a copy has been loaded and to verify that that isn't a false positive.
So if you want to use identical copies of something use instances, those definitely only load the geometry and textures once.
Also I make VN's using iray. I frequently have more than a single person in a scene.
It is rarely the case that I need a crowd for the scene to work, I can't recall one in over a year. Just show the characters that matter in the scene to tell the story. When you do need groups you have to really cut the texture sizes down, a 4k texture for every surface of a background characters armor? How does that make sense? Also watch the subD. Why have a huge geometry figure to show all sorts of small detail on a character 20 meters from the camera?
Getting multiple characters into a scene can be a challenge. If you don't want to go the composite route (which the big studios use) you can use the various utilities available to reduce texture files. One that Mattymanx made is good https://www.daz3d.com/resource-saver-shaders-collection-for-iray
Getting a Titan RTX with 24 GB is a possibility... only a thousand more or so than a 2080 ti.
Remember, a character in the background does not need all the texture data provided. If you do it correctly no one will notice and you can save money.
Games do it all the time, and in real time. How hard it is to include checks of the file pathes of the textures being loaded into memory?
daz studio + iray is a tool for artists or architects not for movie makers or storytellers; when I want to setup a complex scene within a Stonemason set, I always use 3delight or carrara, both are powerful and easy to handle, and the one pass render can be enough realistic to satisfy your needs. This is my last work with a stonemason set rendered in 3delight, only few minutes to get it
Games do not. Games use every cheat that they can come up with to avoid stuff like that. The only time a game would show the exact same texture is if multiple identical objects were shown on screen at the same time, which pretty much never happens. Checking file paths means storing the file path which means creating some additional data structure for every single graphical object to store this piece of information which is only needed for this relatively uncommon occurence that most people simply do not care about and will never ever use.
People already complain bitterly about the length of time it takes to load a scene in DS. If you add a step where every map is checked against the great long list of other maps already loaded loading a scene would take substantially longer.
Game NPC do share textures, otherwise shooters will not be able to generate waves of enemies with piles of dead ones already on the ground. It should be how Daz handles characters but for some reason, based on what I read, Daz instances cannot change poses or else they lose their "sameness".
You need to learn tricks of the trade.
You are taking a LOT for granted. Professionals use billboards, multi-layer renders, level-of-detail functions, and various other tricks to make things seem "populated".
Tools or concepts you would be interested at looking further into...
Decimator: It can reduce models polygon count for creating scene-matched detail objects. I believe it also may reduce textures too, as part of that process. https://www.daz3d.com/decimator-for-daz-studio
Billboards with normals: Rendering a model and capturing the "Normals", from the "Canvases", will allow you to make a light-reactive billboard item. Thus, it will light-up and reflect almost like an actual 3D object, but still be a flat item.
Grouping-renders: Once you have dummy dolls or billboards or decimated models setup, in a large scene... You can part-out the scene in layers. Rendering each layer with HD model replacements and optionally repeating the "billboard+Normals" trick, or just post-work layering of each rendered layer-group.
Blind renders: With all your individual items already "setup"... Models materials, scene materials, lighting... You don't have to work in a live-preview. Work in wire-frame, or simple shaded mode, like most professionals do. Hit "render", only when you are ready to actually render. Being sure that you have adjusted the models geometry and materials to be "reasonable", for the render you are about to do. (No SubD levels of 4 on distant people and hair. No SubD on anything with a flat surface or non-curved surfaces. No 1080x1080 images on EYES on models that render eyes as a single pixel in the distance. No complex shaders or multi-layered images in the EYES, that you can't see in the distance. No caustics, unless they are needed. No "scatter and transmit", unless it is needed. No refraction, if a simple "opacity value" will suffice. No crazy mirror-facing-mirror reflections. Use "object instancing" where possible. Avoid uber-complex things like trees and grass, that can honestly be rendered and added in post-edits, many times.)
At the end of the day, this is an art...
Professionals don't just throw entire worlds into a single computer and expect it to "work"... It takes a lot of computers and teams of people to "Make it work", enough to "trick the audience".
Post edit population... You have an advantage, you can render your assets to perfection. They have to use :stock art" and "hope" they find the right pose and perspective and lighting. You can just make it!
http://www.horoma.net/tutorials/how-to-populate-your-render-seamlessly
Have you actually tested this? It should be fairly simple to do, and there's no point in a long debate on the behaviour without actually verifying that it is the behaviour.
No, they really don't. First that sort of scene is very rare. Second a pile of bodies isn't some sort of soft physics simulated and rendered set of figures that were just active. It's a pre created pile that has no internal surfaces and likely none on the side away from the camera.
No idea what you read but I change poses of instances all the time. Otherwise they would be utterly useless
Well, if instance can be posed then my bad for spreading false information. There is no exact measure within Daz for VRAM usage so I have no way to test if what I read were true as I have only a 4GB GPU and two G8 almost always drops to CPU for me.
I have a friend who did a lot of Daz rendering with a 4Gb 1050ti. You need to use something like Scene Optimizer to get even minimal scenes to fit.
You could load one of a characte and render, noting the memory used, then load a second (or use Edit>Duplicate) and render, and keep going - see how the memory changes.
double post
isn't there a geometry hider script that just came out? can probalby use that and scene optimizer to get very renderable scenes.
Game engines have to reduce geometry and textures by an insane amount to get them to render. 3d Renderers don't really optimize things the same way as game engines. So you wont' be able to render a ton of things in any other renderer either, like Maya, which has been around forever.
For very crowded scenes the only way to render them is by instancing objects (which is sort of like cloning). I rendered this VR scene awhile back with Ultra Scatter Pro.
There are some environments in store that have a lot of objects, but they aren't instanced objects. so they lag more than they should.
For what it's worth: when using two or more identically textured, non-instanced, characters or objects in a scene in 4.10 and 4.11, Scene Optimizer optimizes them separately. That could be due to how the script was written, but it at least suggests that DS loads separate texture resources for each object.
With one figure loaded:
With that figure duplicated (which makes a second copy, not an isntance):
So no, textures are not doubled. There is an increase in the (lower) amount of memory used for materials, though not a doubling.
The behaviour of a third-party script doesn't suggest anything - at least not in that direction (if a script or plug-in broke a supposed limitation that might be more revealing)
according to the image name this took 2 hours in 2015 in iray .. had to have been with a single 980ti6g
11 characters (genesis base) all with mostly different outfits but even where the outfits are the same (like on the paper boys) the textures are different
Hmm, need to do this again with more lighting control.. instead of those god-rayed street lights.
---
I seem to remember that when I had the old machine, (and 3dl) the system was 2g ram, the scene with 9 characters was 20g and I had to sit and keep telling windows that it didn't need to close the program to free up memory because the ram usage of the render would push the ram usage to about 23.5g and windows didn't like having on a half gig for misc.
what set is that?
NO, you cannot change the pose of an instance, independent of the base figure as an instance has no independent rigging.
The only change you can make independent is position.
To change the pose of an instance, you'd have to change the pose of the base figure.
I do things like open drawers on instanced end tables without issue. I have no idea what you mean by poses.
Do you notice the two figures have the same exact pose, right arm up and left hand to the left hip? The instanced figure's pose is controlled by the base figure. Changing the pose on the base figure automatically changes the pose on the instanced figure to the same pose. Their poses will always be identical. Does that help?
one that will be available once I get my skills honed enough to condense some very complex groups of objects down to simpler ones. A lot of the stuff you see is still primitives that haven't been converted into single meshes.
I started building using methods I learned in construction and set building for the theater.. now I"m working on 3d skills also. Good training for that is https://www.daz3d.com/the-abc-s-of-3d-modeling on sale right now for $10
But that's a set I build based on a Norman Rockwell painting. I do have some Hopper buildings also. Working on a street system with lights etc that you can shoe horn in the buildings to where you want them
---
the picture with the building 2kx1k only took 47 minutes with the tititanX and used 11800 of the 12200 space on the card.
---
the other one is an overview of the blocks system
---
*and no, I didn't do the cars... nor the sky which is one of Dimension Theorys skies of iradience But the rest I'm guilty of
No, because you seem to mean moving bones and you can certainly do that.
Super work!
The only way you can open that drawer is either the drawer is an instance itself or you are manipuating the original object.
To clarify a bit here, you cannot directly manipulate what is a sub-node of the original figure in an instance, unless you manipulate the sub-node of the original figure.
That is, if you instance G8f, you can't move the instance's big toe, you have to manipulate the original's big toe, changing the position of both.
You can however, change the base nodes xyz translation and rotation as well as scaling the instance.
I hate that so many vendors are making these huge mega scenes that are so power and time hungry on the computer and virtually take days to render. I wish that there were more individual item sets available as far as props that can be arranged yourself instead of having one massive set that I can never use because render times are ridiculously long. I do know that I can use items seperately from scenes but I am still faced with hardrive filling up with a lot of scenes that I will never use.