Faking it w/ BI


Blender Internal
render time ~ 3 minutes @ 800x600 on Q6600 quad core (2.6GHz), 8GB 800MHz DDR2 RAM, WinXP x64, 64-bit Blender 2.5Beta, 4 threads

what exactly are you trying to fake?

Isn’t it obvious? The red ball is fake bacon. HERESY!

Ahh… how foolish of me :stuck_out_tongue:

Maybe trying to get Luxrender/Yafaray level realism using cheats in BI (which this attempt doesn’t seem to get real close)

I’m impressed by what you do have, though I might suggest a simpler scene to focus more on the faked caustics.
also the glass ball on the right… assuming that’s what it should be… need to be more transparent to be making that bright of a spot, just my thought though.

Your textures are very bad. You can see the “confetti” discoloration from the jpeg compression on the wood, and the tiles lack 3d depth (you really should just model those).

If you want a really nice wood texture, here’s what you do: Photoshop, make a new image as big as you can while still being a power of 2 (16384x16384) because in photoshop the clouds effect tiles when image dimensions are powers of 2. Then scale it down to 2048x2048 (or 512x512, whatever you want) and add a long vertical motion blur to it. Now you have blurry clouds, use hue/sat/lum colorize to make it a light brownish color. VOILA! Wood texture!

It’s one of the few things that’s easier in GIMP, just make the clouds smaller instead of making a huge image and scaling it down (curse you Photoshop and your lack of cloud effect controls).

Or alternatively just make a procedural material using either a cloud texture, or the wood texture, or a combination of the two and possibly others depending on the look you are going for.

Sorry, I didn’t have time to post exactly what I was after when I posted the image… was hoping that someone could figure that out (like Ace Dragon did) until I had a chance to come back and “do some ‘splainin’”.

I am not new to 3D, but relatively new to Blender.

The textures, I 'm not concerned about… they are “low res” for a reason (the low camera angle and and the re-use of the same texture throughout shows how Blender filters the texture for far away (baseboard), mid distance (chairs), and close (table).

This is a simple test scene of various lighting and materials setup in Blender to try and get BI to render something close to what is possible in something like Yafaray/POVray without the horrendous render times. (I only added the chair to add something a little more geometrically complex than the walls and the table/cloth) and the tablecloth (which replaced a pre-existing attempt by “others”)… the rest of the “room” was provided by “others” as they were trying to use POV-ray but found the render times w/ caustics and reflections to be too long for animation.

I told them that I may be able to help get them setup with something in Blender that rendered nearly as good, but MUCH faster.

I did UV map all the objects where required and I set up the materials… some more successfully than others. The ubiquitous chrome & marble balls turned out OK, I think, but I’m not happy with the red glass at all… but that’s why I posted here…

I am also trying to fake caustics using small spot lamps w/ custom falloff curves… I think the red glass caustic is a bit more successful than the chrome caustic here…

I am looking for any tips/tricks to:

  1. improve the look of the glass, ray traced transparency is being used.
  2. improve the look of the AO (the darkened corners I couldn’t get to go away, or lessen at all… I need help there!
  3. I’d like to get rid of the area light but I couldn’t get AO to darken the floor below the table… without the area light, this scene renders in just over 50 seconds… so, some help there would be much appreciated.

Things I already know:

  1. should have put the fake caustic spots, the balls, and the plate on a separate layer and used the “this layer only” light option to avoid the light bleeding onto the tablecloth as seen below the plate just below the chrome ball.
  2. as stated before, texture resolution needs to be higher, or moved to a procedural for shots close to the camera to avoid pixelation (or increase the filter size, I guess).

This is not art… this is a TEST. :wink:

Sorry I wasn’t clear before…

I’ll ask the “others” if it is OK to publish the .blend… if it is OK, than I can send it to anyone who is willing to take a look and give me some tips.

improve the look of the AO (the darkened corners I couldn’t get to go away, or lessen at all… I need help there!

Default AO max_distance is 5.00. Increase it to 20 or 40. But it will remain there somehow to keep you unhappy LOL. So lower AO and use some more area lights instead.

Another trick here is to use a DOF map as a multiplier, so more light in front more dark background. Use nodes for this.

area lights are a problem… I have just one in the scene and it drives the render times up pretty badly (from 50 sec w/out to over 3 min. with) so I’d like to avoid them if possible. The idea is to get a decent GI look with faked caustics while keeping the per-frame render times below 3 minutes for animation. Hopefully, the increased AO distance in conjunction with your other suggestion (node based DOF map) I can limit the adverse effects of the increased AO distance. :slight_smile:

I have to spend some more time with the compositor and node based materials. I know they’re there, but have only limited experience with them in Blender.


especially this chart

See how the input node opens more keys.
Read these two links. Maybe difficult to find as wiki help of blender is a nightmare. All this community, all of us should be ashamed for this. Sorry for my english.

If you’re having trouble with AO may be you’ll want to give this (pretty old) trick a try:
(see figure 12.50).

Thanks for the links guys!

hehe… looks like I have a lot of reading to do!
But it’s overdue… multiple passes + compositor is really the only way to get good controllable, easily tweak-able renders.

I’ve used the “light dome” method before in other apps, but I don’t think it’s the right solution for this particular scene, since there is only one visible overhead light source. However, I will experiment with the area light replacement in the 12.4x section of that link (spot “dupliverted” to a sub-divided plane, as opposed to an icosphere ).

I’ll have you know that I am truly offended that you would think I would “fake” bacon.
Bacon (and beef jerky) is sacred in my world and I would NEVER stoop to such.
Bacon (and beef jerky)… the REAL kind… just mean too much to me in my life.

Anejo… I have that word printed on the several empty Patron bottles that clutter my desk right now! :o


All other comments also applying, the biggest problem I can see is the lack of “lighting depth.” The light in the scene is cast so evenly on all surfaces that it flattens the depth impression despite some obvious depth clues like perspective and overlaps, and gives the image the look of a poorly-executed trompe l’oeil. Even subtle light fall-off could not only give better focus to the composition, but also improve the sense of spatial separation in the various picture planes.

Your glass(?) ball has some reflection and/or refraction issues. If intended to be a transparent glasslike material, then the reflections are much too pronounced, looking more like refractive effects, but which would be impossible given the composition – the red ball would have to be behind the glass ball to be refracted a shown. If a highly reflective material, then the caustics are out of place, given the implied overhead light source – they disobey the laws of reflection.

“Lighting depth” is an issue that I will address in a scene by scene basis, but I will be testing the above mentioned techniques (render layers, DOF layer as multiplier, etc.) on this scene to achieve the look I’m after. The lighting is currently very simple… (1) overhead area light, and (2) fake caustic spots. Lack of depth with that setup is pretty much expected, right?

I’m not really looking for tips on composition, more on techniques that can be tweaked (hopefully relatively quickly) from scene to scene. I mean… 3 balls on a plate on a table in a nearly bare room is hardly the subject of “high art”. :wink: This is barely one step above a chrome sphere over a checkerboard plane! :slight_smile:

The fake caustic light on the reflective/chrome ball is definitely wrong… but I was trying (unsuccessfully, I know :wink: ) to achieve something that I CAN see in a render of a trimmed down version of this scene in the Kerkythea render. Seems like there is a caustic reflection of the the plate’s light bounce on the near side of the chrome ball. It’s nowhere near as focused as in my image, and it’s way more spread out, roughly from just to the left of the red glass ball, all the way around the inner bevel of the plate, to just to the right of the chrome ball. But it’s so unexpected with the apparent overhead light source (the box in the ceiling has an emit value in blender but is actually a mesh light in Kerky), that I’ll likely get rid of it altogether to further decrease render time.

Thanks for the comments, chipmasque.

I remember your thread for your Kata model over in the Blender section on CGTalk. I remember a certain user called “Oddity” givin’ ya guff over it… some prudish crap about “tacky and uncouth smut” instilling in him “an aesthetic and intellectual disgust”. what a hoot!

Do you, by any chance, know what ended up happening with that guy?

Couldn’t care less :rolleyes: :wink:

My comment about the lighting was intended to show what seems to be a weakness in the BI “interpretation” you’ve made – hard to say how it compares it an unshown image from another renderer, but one aspect of many of the external rendering options is that they usually deal with light within a volume, meaning that they allow for fall-off within the scene space, as well as bounce and other factors. It’s as much a technical observation as an aesthetic one. BI lighting can do a similar look but it takes more finessing of the light sources, in my experience at least.

Now that I know the ball is supposed to be chrome (I had missed the earlier reference on scanning the posts), and how you describe it looking in the other renderer, I can see your intent, but the caustic pattern looks to me much more like something a glass sphere would produce by refraction than a chrome sphere would by reflection, hence my confusion in that regard.

Given his attitude and comments on your work, this link ought to give ya a good laugh if you enjoy irony:


It’s tips on this “finessing” that I’m looking for. It seems everything I try that makes teh lighting better, does so at a very high cost in terms of render times.

Yes, I can see that. Much too hot at the point of contact. Reflective caustics tend to be projected further away and have a distinct shape to them that would be impossible to achieve with a simple custom falloff curve. Another reason to drop the attempt entirely for this test.

You mention not liking area lights due to the rendering overhead they produce, but they seem to me to be the most versatile for producing certain kinds of lighting effects. I’m very partial to them, though I agree that they can introduce massive render times depending on the subject matter – for instance, some close-ups in Kata require from 30 min to 1.5 hr/frame depending on the machine (I use a three-comp mini-farm nowadays), due to raytracing the particle hair, But the look I want can’t be obtained any other way so I bite the bullet and plan on extensive rendering sessions for those scenes.

I’ve found there is always a need to compromise between a desired “look” and the practicalities of rendering animations, because you can only “fake” it so far. The compromises also seem to be pretty scene-specific; I’ve rarely found a situation where a single lighting solution can be applied to many different scene setups, even when the subject matter itself isn’t changing much.