Karl's Journey

Another topic that comes up often in the Daz Forums is adding backgrounds to 3D images. While I wrote up a quick guide to my views on the ways to do this in the Question about skydomes and other types of backdrops thread, I thought I would expand a bit here to discuss what I see and the strengths and weaknesses of the various options. I link to a few images where I have used these techniques

HDRI images

HDRI Images are easy to use in Iray rendering (and not terribly difficult anymore in 3Delight) and probably the most common way right now to add a sky or background to your image. They have a few advantages in that they completely surround your scene, provide lighting which automatically matches your background (no more shadows going left in the background when the lighting on your figure casts a shadow to the right), and often include a ground texture for a full environment. The drawback is that the quality of the HDRI will drastically affect your experience with them and you are somewhat limited in how you place your figures and props within them since you can’t move the dome. Also they have to specifically be designed to support Depth of Focus (DOF) from your camera; so most of them will be sharp and in focus despite looking like the background is far away from your figure.

If you’re going to use HDRIs from a source like HDRI Haven you will do yourself a favor to get used to the options for either Infinite or Finite sized HDRI domes. Also tools like the IBL Master and Iray HDRI Toolkit will help you use HDRIs more easily and get more value from them.

Images like Neko’s Lazy Sunday and Jogging in the Park are only using HDRI images for the backgrounds.

Backdrop

If you want to use the Backdrop feature (Environment Tab (not to be confused with the Environment tab in Render settings) – select an image for the background instead of a color), you can do it, you just need to edit the image you’re going to use to be the size of your render. That will avoid any distortion of the image when Studio tries to make it fit the background.

I used that technique for my Neko in Venice image. The main advantage to the backdrop over the flat primitive is that it is technically “outside” your scene; so you don’t have to worry about it blocking lights. It also “follows” your camera so you can change camera angles to frame it the way you’d like without having to move the plane behind it.

The drawback is that if you change your mind about how you want to frame the image and it affects the cropping or aspect ratio of the background, you’ll have to re-edit the background image to map to the new dimensions.

Flat Primitive with an Image

For flat backdrops, when you have a specific camera angle already setup, using a primitive plane and putting the image on there is the most flexible. You can size the plane so that the image has the proper proportions and move it around to use only the portion of the background that you want in your render.

The drawback is that you have to pay attention that it isn’t getting in the way of any lights or props you have in your scene. Also, if you move your camera, you may need to re-position the plane to align with the new camera angle.

Curved Primitive with an Image

Several environment sets for sale use these. The advantage over the flat primitive is that the background curves around your set so you aren’t limited to only shooting in one direction. This is particularly useful when you’re shooting an indoor scene and want to have something outside the windows but don’t want to be limited in your camera angles.

While it is more flexible to camera angle adjustments than the flat plane is, you could still run into cases where moving your camera means you need to move the backdrop as well.

Skydomes

By this I mean the “old” way we used to do this which was to create a large half-sphere which covers your entire scene and has an image mapped on the inside. Advantages here are that I get the flexible camera angles like the curved primative, I can move and re-size the dome to adjust to my scene unlike HDRIs, and I get the full sky and surrounding environment. Drawbacks however are that I won’t get a ground like with most HDRIs.

Also, these can cause issues when you’re rendering with Iray. For instance, if you set your Render Settings / Environment to use Sun-Sky or load an HDRI sky image thinking it will provide daylight, technically Iray will think there’s a solid dome covering everything; so the light you’re expecting to come from the Iray environment settings will be blocked by your skydome.

This is one of the more common reasons we see people coming to the forums and asking why their lighting isn’t working on an older environment set.

Cylcloramas

For example, the “Millenium Environment” that comes with Studio. These are really falling out of favor. They have an advantage of providing true depth where some elements are closer to the camera than others. And they are lightweight environments on your computer because all of the background items are pre-rendered flat props instead of full 3D geometry, but you have to be REALLY careful about lighting and camera angles or the illusion is spoiled; so they aren’t very flexible.

No Backdrop

Finally, I’d be remiss if I didn’t also mention that many people will choose to render without a backdrop at all and composite their render over a 2D image using something like GIMP or Photoshop. Since those 2D image editors provide other tools as well for postwork, this can be a great option and is a technique I used for my Walking a New Path image in the gallery. If you’re doing this, you’ll probably want to learn a little about Iray Shadow Catchers because casting realistic shadows on the backdrop is the trickiest part to this technique.

Summary

There isn’t necessarily any “right” or “wrong” to any of these options. They are choices that you make as the artist to what fits best with the image you’re trying to create. I’ve given what I see as the strengths and weaknesses, but you will have to experiment a little and choose for yourself.

I’ve typed this post multiple times on the official Daz 3D Forums; so I thought I’d write it up here so I can just refer to it instead. Maybe this will help some people or at least give you some ideas about how to create your own workflow that works better for you.

My Scene Creation Workflow

As for my overall workflow, I tend to build scenes using this process. Sometimes I short cut it if I’m doing a quick scene “just for fun”, but anytime I want a quality result, I use this process.

1. Scene Blocking
   1. First I’ll figure out the setting (environment) and rough camera angle I want to use. I make sure to use an actual camera and not just the perspective view or so I don’t lose the blocking of the image.
   2. I may also rough in the lighting at this time like adding HDRI skies and other lights I think I’ll need.
   3. I save this as a scene, and start a new one.
2. Character Setup
   1. I load one of the gray background HDRI presets from Colm Jackson‘s PRO-Studio HDR Lighting package. I like the even, low-contrast lighting for character setup.
   2. I load my first character I’m going to use, tweak skin, eyes, add hair and clothes.
   3. A recent addition to my workflow is to add Canary’s Cameras as they provide a quick way to look at my character setup from multiple angles. Saves me a lot of time.
   4. I will do several test renders to make sure things look okay rendered.
   5. If the character is going to pose by themselves in the scene, I may choose a pose for them at this stage. If they will be part of a “couples” pose, then I leave them in default.
   6. I save this as a “Scene Subset” rather than a “Scene” because the Subset won’t include the HDRI I loaded for the setup or any cameras I created for test renders. (Refer to the note about Scene Assets below for an alternative for frequently used characters.)
   7. I repeat the above for each character in the scene.
3. Optional – Couples Posing
   1. If the characters are going to be posed using a couples pose, I’ll create a new “character setup” scene with my standard HDRI
   2. Then I import the Scene Subsets for each character and apply the couples pose to them.
   3. I will adjust the poses if necessary, doing some quick renders to make sure that the intersection of characters, clothing, etc. looks correct. V3Digitimes‘ Ultimate Pose Master has made this process SO much easier and quicker for me!
   4. Then I create a group with the two characters and save this as a new Scene Subset, excluding any cameras and lights I added.
4. Character Placement
   1. Now I’ll reload the original scene and merge the character and/or couple scene subsets into the main scene and move the characters into their positions in the scene
   2. This is where the couples group comes in handy because I can move them together without having to adjust each one individually
   3. Now I’m tweaking poses, adding or removing other props, etc. to get the final look for the image I want.
   4. During this time, I use the Aux viewport set to my main camera view while I’m using the Perspective View to move pieces, adjust poses, etc. That way I can see how the changes look from the main view.
   5. I also swap from Perspective to Main Camera on my full viewport to make sure I’m getting the results that I’m trying to and that I’m not spending too much time on something that won’t even show in the final render. I can’t tell you how much time I wasted trying to get a hand to be in the correct place only to realize it was blocked from the camera anyway and nobody would notice all the time I spent getting the fingers to just lightly touch the other character’s waist.
5. Lighting
   1. Once everyone is in place, I start tweaking my final lighting, doing test renders and adding Ghost Lights, etc. to get it looking correct
   2. With some renders, this is the longest part of my process as I feel the lighting has the greatest impact on the overall quality of my images
6. Final Render
   1. I always do my final renders using Iray canvases, even if I only intend to use the Beauty pass. Canvases contain richer information about the image and so are a better basis for any post processing I want to do.
   2. For higher quality images, I’ll also create canvases for each light source so that I can adjust their strength in the postwork
   3. I also always render about 20-50% larger than I want the final image to be. That allows me to downsample the image post-render which can help reduce any remaining “noise” in the image.
   4. If my goal is the highest quality I can produce, I’ll disable Iray Tone Mapping and set the render quality settings to 100% convergence, Quality factor of 8, and both max samples and max time constraints to max values to allow the render to go as long as necessary.
7. Postwork
   1. I use GIMP to import the EXR files from the Iray canvases and perform exposure adjustments, layering the light layers, etc. I save this as a GIMP file so I can come back to it if necessary.
   2. I will export the image as a PNG in it’s full size, then open the exported file and resize it. This way I don’t mess up the post process file I saved in the previous step.
   3. Even though I calibrate my monitors regularly, I’ve found that mobile devices display them much differently; so I’ll open the final file on my mobile phone and use Adobe Lightroom to perform final tweaks to the exposure and color balance settings before I post it anywhere.

An Aside about Support Assets

For frequently used characters, consider using the “Save As -> Support Asset -> Scene Asset”. This has a value over saving a scene or scene subset in that updates to the components of the scene are reflected in scenes that use them. To put it more clearly, if you have a character you use often that is setup using Aiko 8 with OOT’s Linda Ponytail hair. You get her basic setup and save her as a Scene Asset (let’s call her Aikolinda). Then you load Aikolinda into one (or more) scenes you’re going to render. Then you decide you don’t like her as a blonde and want her to be a redhead. If you go back and update Aikolinda with the new hair color, that change will automatically be reflected in every scene you loaded her into. If you had saved her as a Scene or Scene Subset, you’d have to apply the hair color change manually to every scene you used her in.