PBR VS Raytracing

[Chiefer]

On 2/27/2020 at 3:31 PM, Designer1 said:

 

 

I also noticed that even though I have the smooth lines when idle on some of the lines seem jagged and not as smooth as raytracing. 

I not sure why this is happening to you but turning on Edge Smoothing eliminates the rough edges...

[Designer1]

Graham, thanks for the additional information I will give that a try.

 

Jintu, yes I had the edge smoothing turned on so that's why I was unsure why this was happening.  Maybe its like what Graham said that if the camera is slightly off angle it can cause the lines to be jagged.  I will try a few things out and see if it persists.

 

Thanks for the all the help!

[TheKitchenAbode]

Here is an example of the jaggy/tearing. Note the left side of each window casing and the direct light outline of the sun cast on the wall.

 

Camera Tilt = 0 Degrees.

 

Camera Tilt = -2 Degrees.

[Designer1]

Wow interesting Graham thanks for the examples.  Were both of those PBR's done with the edge smoothing when idle on? I didn't think such a small angle would create this.  Thanks for letting me know, I will make sure this is taken care of before taking the picture.

[TheKitchenAbode]

3 hours ago, Designer1 said:

Wow interesting Graham thanks for the examples.  Were both of those PBR's done with the edge smoothing when idle on? I didn't think such a small angle would create this.  Thanks for letting me know, I will make sure this is taken care of before taking the picture.

 

Yes, I always have edge smoothing on. Just be aware that this does not happen every time the camera is tilted, it all depends upon the suns intensity in conjunction with any given tilt. So if you wish to have the tilt but are experiencing some jaggies then try increasing or decreasing the suns intensity, just might do the trick.

[Alaskan_Son]

10 hours ago, Designer1 said:

I didn't think such a small angle would create this.

 

It's not that such a small angle creates the problem, its that the problem only comes into existence at a specific combination of angles and it only takes a single degree to put it over that threshold.  This is true of almost every software limitation.  There's a specific threshold.  As I mentioned earlier in the thread, this same problem occurs with reflections in PBR's too.  It can get really frustrating because you're essentially forced to position the camera where you don't actually want in in order to get acceptable shadows and reflections.  Its one of the big areas where I see Ray Traces remaining irrefutably superior.    

[Chiefer]

6 hours ago, Alaskan_Son said:

 

It's not that such a small angle creates the problem, its that the problem only comes into existence at a specific combination of angles and it only takes a single degree to put it over that threshold.  This is true of almost every software limitation.  There's a specific threshold.  As I mentioned earlier in the thread, this same problem occurs with reflections in PBR's too.  It can get really frustrating because you're essentially forced to position the camera where you don't actually want in in order to get acceptable shadows and reflections.  Its one of the big areas where I see Ray Traces remaining irrefutably superior.

Correct, and it's not the tilt.

[Designer1]

Thanks for the information everyone. Yes I can see what your talking about Michael with the shadows.  I did a PBR and the linen tower appeared to have a shiny surface, even though it wasn't set to be shiny yet when I selected matte the texture looked out of place, like I just used clip art and placed it into the bathroom, it was strange. 

 

The time it takes for raytracing is still so much that's why Im trying to use PBR.  Sometimes for a really crisp and clear raytrace it can take 20-30 minutes... so if you have 100 raytraces to do it takes forever.

 

Glad at least that we have two options now to try and work with.

[TheKitchenAbode]

Chad, transitioning from CA's Raytracing to PBR'ing can be challenging as each rendering engine interprets things differently. Sun, Lights and materials in most cases need to be adjusted according to the chosen rendering engine. The example below, though extreme, demonstrates this. The first render is a PBR, lighting and materials adjusted accordingly. I sent the scene to Raytrace and as you can see it is distinctly different. Please do not interpret this as demonstrating that PBR is better than Raytrace, it's just a demonstration that things need to be adjusted to best suit the rendering method you wish to work with.

 

PBR 1200 X 600px

 

Raytrace 1200 X 600px, 40 passes, 30 minutes.

 

Regardless of which CA rendering type you use, neither is perfect, they each have their deficiencies. I can only speak for myself but after spending some time with CA's PBR I can produce renderings that meet my needs far faster and of higher consistency than I was able to achieve using CA's Raytrace.

[Donco15]

Graham

Great Renders showcasing both methods.. Thanks for showcasing this to show how the same image looks completely different with the two methods.  Amazing... This is so helpful to see them side by side. The PBR version has more warmth and brings you into the photo.. This is my fav of the two... Thanks again for sharing...

[TheKitchenAbode]

11 minutes ago, Donco15 said:

Graham

Great Renders showcasing both methods.. Thanks for showcasing this to show how the same image looks completely different with the two methods.  Amazing... This is so helpful to see them side by side. The PBR version has more warmth and brings you into the photo.. This is my fav of the two... Thanks again for sharing...

 

I'm certain that a Raytrace could also be done that also has more warmth, it's just that everything would need to be adjusted to generate that look.

 

The biggest issue I have with CA's Raytrace is the time it takes to render, for every material or lighting change you need to run a trace to see what the effect is. In my example I had to run the Raytrace for at least 10 minutes for some of the adjustments I needed to do so my PBR would at least Raytrace. Keep in mind that the 10 minutes was for a Raytrace at only 1200 X 600 px, a larger sized scene would have taken much longer. With PBR you can work in it live so you see the effect right away and pic size does not really affect the rendering time.

 

I'm working right now with a client overseas, they e-mail me a change and I e-mail them back 1 or two minutes later with an updated rendering. No way could this be done using Raytrace.

[Designer1]

Graham, great examples, thank you!  Yeah Im noticing everything you have mentioned on this post about the differences between raytrace and PBR.  I noticed on interior PBR's that the more you reduce the suns lumens the brighter the interior lights get and the harder it is to control them without it getting dark.  So its been a balancing act.  I LOVE raytrace shadows and wish PBR had them especially for the landscaping.  Your right though the amount of time it requires is too much.

 

My biggest frustration at this point is reducing the shadows of the PBR trees.  I can reduce the suns power down to 1000 and add a light of 1,000,000 lumens and it still doesn't reduce the amount of darkness in the shadow very much.  So clearly im not selecting something in my camera DBX or am missing something completely.  Can you tell me on the post you were showing your lightened tree shadows how many lights and at what lumens they were to get the tree shadows to look that good?  I cant get even close to that... they still look sharp edged and fake.

 

Thanks!

[TheKitchenAbode]

Chad - on your added light do you have shadows turned off in the lights DBX? Turn shadows off and this added light acts to provide ambient light to the scene, the sun is the direct light from which the shadows result from, the ambient light injects light into the shaded areas and brightens them up. Keep in mind with this shadow thing that in real life the sun does not directly create shadows, shadows are actually regions lacking in light, the light from the sun is being blocked by objects. The issue we have is that in real life the light coming from the sun, though it is direct light at it's source, is altered as it hits the atmosphere and other surfaces, this creates a degree of scattered light so what we see is a mix of direct and scattered light. It's this scattered light that lightens up the darker areas. In CA we have no direct control on the sun to define the ratio of direct versus scattered light, so we need to add another light source to simulate the missing scattered light. When I add this light I treat it as another type of sun and place it up in the sky, maybe 300" or 600" above elevation 0 and in approximation to the suns position relative to the model and with shadows turned off. Essentially I have two suns, one that produces direct light(CA's sun) and another that simulates scattered light(the added light), adjusting their intensity levels(ratio) will allow you to control the shadows degree of darkness(lack of light). When doing this remember that CA adds the light contribution of each together so as you increase the added lights intensity the shadows depth will reduce but the overall scene will brighten. This needs to be countered by reducing the suns intensity to restore the desired overall scenes brightness. Once you get things close enough you can always fine tune the overall brightness with the PBR Exposure and Brightness sliders.

[TheKitchenAbode]

12 hours ago, Designer1 said:

I noticed on interior PBR's that the more you reduce the suns lumens the brighter the interior lights get and the harder it is to control them without it getting dark.  So its been a balancing act.

 

Yes, with CA's PBR the sun affects the overall interior ambient light level, this is also impacted by the % of glazing. There several ways to bring back the ambient light level, one as I previously mentioned is to add a very minor amount of emissivity to the wall and ceiling surfaces. Another is to use area point lights, usually at low intensity levels with several spread around the room. The other is to keep in mind that you can add additional light sources to existing light fixtures. For example, a default recessed light fixture has one light source, a spot light. You could add an additional light source to this with different settings than the default one to take care of other lighting needs such as shadow control and ambient light contribution. The thing is that in CA each type of light source, either spot or point, is limited in what you can adjust so it's virtually impossible to replicate(simulate) how a real light bulb actually works with only one light source.

[Designer1]

Thanks Graham!

 

Yes, the interior PBRs were helped greatly by your recommendation of the emissivity of the walls thanks for that idea!  As for the added lights for the exterior landscaping I believe I had the shadows turned off.  I do know that my fake sun (added lights) were very close to the ground because I thought if I had it too far away from the trees shadow it wouldnt lighten it up.... so maybe if I place it, like your recommended 300-600' above grade 0 maybe thats the trick.  I will give a few of these ideas a try and see how it works out.

 

I find the PBRs to be like a balancing act and you cant really use common sense when trying to accomplish a good realistic PBR.  Its like since everything from shadows and sun are fake and you dont use real life lighting logistics to create a good rendering, you have to change the way you build the landscape and scene.  I found the raytraces, although more time consuming, to kind of automatically take care of some of these technicalities that you have to consider before starting the PBRs.

 

It will be interesting to see in future releases of chief how they further develop the raytraces and PBRs.

 

Thanks again!

[TheKitchenAbode]

Chad - you can place the exterior light(s) locally near the shaded area if you wish. I just find that it's a b it more trying to get things right and often requires more lights, you would need to do this for every shaded are you need to reduce. Using one light placed up in the sky with shadows turned off will affect all shaded areas. When you turn off a lights shadow what you are doing is telling the light to ignore anything in it's way that would block the light from traveling any further. The light lights the objects exposed surface and then carries on to the next exposed surface and lights it, this continues on and on.

 

As far as balancing things out and the underlying logic behind PBR, I find this to be fairly aligned to how light actually works. In respect to CA's Raytrace the biggest difference in my opinion concerns the sun, in Raytrace the sun is treated as a direct light only while in PBR the sun contributes both direct and indirect light. This makes a significant difference in interior renderings. What I dislike is the fact that there is no option in PBR to control the degree(mix) of direct and indirect light, it's fixed at some predetermined ratio. There really needs to be a mixer slider that allows us to adjust this for any given sun intensity level. Such a slider would also provide shadow control as raising the % of ambient would reduce sun related shadow depth.

 

It does take some time and experimentation to get a handle on CA's PBR, bit now having done a lot of that I would certainly never go back to CA's Raytrace renderer. It's not perfect by any means but personally it's the better of the two choices CA give us.

[Alaskan_Son]

2 hours ago, TheKitchenAbode said:

As far as balancing things out and the underlying logic behind PBR, I find this to be fairly aligned to how light actually works.

 

This is a surprising statement.  I couldn't disagree more.  Chief's PBR treats lights somewhat accurately as it appears in  and affects adjacent rooms, but in the current room, it always uses some predefined baseline and builds a sort of aggregate lighting effect around that.  This is easily demonstrated by placing 2 basic rooms side by side.  Place a window in each room, and place a door between the rooms.  Now place a 100 lumen light fixture in one room and a 2,000,000 lumen fixture in the other.  Set up a camera in both rooms and then experiment a little...

-Toggle the sun on and off

-Close the door between the rooms

-Open the door between the rooms

-Turn lights off in the rooms one at a time

 

Now change that diving wall to an invisible wall.

 

I think what you'll find is that the way both the sunlight and lighting from adjacent rooms (or lack thereof) affects its neighboring rooms is somewhat accurate, but the lighting in the room your camera is placed in is never right and definitely does not behave like light actually works at all.  My hats off to you though for the amount of time you've invested manipulating things to get your desired results. 

[TheKitchenAbode]

3 hours ago, Alaskan_Son said:

 

This is a surprising statement.  I couldn't disagree more.  Chief's PBR treats lights somewhat accurately as it appears in  and affects adjacent rooms, but in the current room, it always uses some predefined baseline and builds a sort of aggregate lighting effect around that.  This is easily demonstrated by placing 2 basic rooms side by side.  Place a window in each room, and place a door between the rooms.  Now place a 100 lumen light fixture in one room and a 2,000,000 lumen fixture in the other.  Set up a camera in both rooms and then experiment a little...

-Toggle the sun on and off

-Close the door between the rooms

-Open the door between the rooms

-Turn lights off in the rooms one at a time

 

Now change that diving wall to an invisible wall.

 

I think what you'll find is that the way both the sunlight and lighting from adjacent rooms (or lack thereof) affects its neighboring rooms is somewhat accurate, but the lighting in the room your camera is placed in is never right and definitely does not behave like light actually works at all.  My hats off to you though for the amount of time you've invested manipulating things to get your desired results. 

 

That's an extreme example, but it does demonstrate an often mentioned pet peeve I have concerning CA's approach concerning Camera Exposure. This exposure thing of theirs does not work the same as it does on a real camera. From what I can deduce, it seams to analyze the light being received by the camera, it then establishes a max pixel brightness and then adjusts everything else relative to it. As you move the camera around, the receiving light changes and the camera exposure thing kicks in and adjusts accordingly. That's why I often recommend setting the camera exposure to max., though it does not completely disengage this it does reduce its effect and it also forces one to use much lower light intensities which also helps the situation. With PBR, if you change the camera, say from one room to the other, you will likely need to adjust the PBR exposure or brightness to correct the rooms light intensity of the room you are viewing from. This is not really any different than a real camera under similar circumstances, you would stop the camera up or down, increase or decrease the shutter speed or adjust the ISO so the room you are in would not be over or under exposed. Other visible rooms beyond the cameras room would be affected accordingly.

 

Just out of interest, one room has a light at 100 lumens, the other at 2,000,000 lumens as suggested. Placed a camera in each and ran a PBR and a Raytrace. Each scene is the same pixel resolution 1200 X 600. The Raytrace ran for 25 passes. Now I think this is an overly extreme test however the Raytrace 2nd pic. could not handle it, light bleed on the shared wall even though I have a roof and foundation on the test house. In the PBR ones I could instantly tweak the Exposure and Brightness to fine tune the scene, in the raytrace any adjustment required running the Raytrace again. Also, even after 25 passes the Raytraces exhibit visible noise, more passes would have been required to clean this up. Raytrace did a better job of exhibiting reflectivity in the glass windows, though in the bright room 4th pic it's overdone and the window looks more like a mirror. Note the door handles and shading on the doors, Raytrace struggles with the brushed aluminum handles. The Raytraced scene also exhibits jaggy/tearing in regions of very high contrast, note bottom of doors and baseboard in contact with flooring.

 

As I have stated many times, neither CA render method is perfect and each requires the user to tweak the materials and lights in order to get the better out of the chosen render method. Ultimately it's up to the end user to decide which works best for them.

 

PBR - Darker room looking towards/into Lighter room.

 

Raytrace - Darker room looking towards/into Lighter room.

 

PBR - Light room looking towards/into Darker room.

 

Raytrace - Light room looking towards/into Darker room.

[Designer1]

Graham,

 

Thanks for all the additional examples.  Im finding your points and descriptions very good, better then watching some of the chief videos.  I think coming from a users standpoint you use the PBR regularly and have clearly mastered the tool to get the desired results.  I will try using your method of lights higher up and add a few.  I think your right about adding a light per shaded tree, that would be hard to get it to look very even given the size of the tree and the shadow.

 

Thanks for the additional examples.  I feel like were all at a great tutorial!

[TheKitchenAbode]

3 hours ago, Designer1 said:

Graham,

 

Thanks for all the additional examples.  Im finding your points and descriptions very good, better then watching some of the chief videos.  I think coming from a users standpoint you use the PBR regularly and have clearly mastered the tool to get the desired results.  I will try using your method of lights higher up and add a few.  I think your right about adding a light per shaded tree, that would be hard to get it to look very even given the size of the tree and the shadow.

 

Thanks for the additional examples.  I feel like were all at a great tutorial!

 

Pleased you are finding this thread of interest. As you can see, there are differing opinions as to what is going on in PBR so you will need to decide for yourself as to which direction to go. I believe there is somewhat of an opinion that I'm using non traditional methods to obtain the PBR results as demonstrated in some of the renders I post. Please be assured that this is not the case. I use the generic sun in a fixed position, adjust intensity and maybe a slight colour. My primary interior lighting is a recessed light that has one additional light source added to provide some shadow control, intensity, angle and drop rate are adjusted. In most scenes these are the only source of lighting, I rarely use area lights and when used they are usually to deal with exterior shadow control. Materials are standard CA, most adjustments are just roughness, color and diffusion. I do not make special material exclusive maps, if I need some bumping I just copy the texture map and paste into the bump map box. Most models are just CA library items and a few are from 3D warehouse, the only alteration I might make is the angle degree in the symbol DBX. That's about it, no magic, no smoke and mirrors.

 

The only problematic material when PBR'ing is glass. The CA default glass material and property definition does not work, both Rene and I posted a year or more ago a few material alternatives that work fairly well, suggest searching the forum and uploading these. They do at times need some adjustment depending upon your lighting, usually just changing the texture size and rotation angle will do the trick.

 

The biggest challenge is getting everything properly balanced and when things do not look as expected understanding what needs to be adjusted, is it the lighting or the material. My recommendation has always been to ignore materials, get the overall lighting correct/balanced and then adjust the materials so they look as you want them to under those lighting conditions. Keep in mind that lighting has a more global effect, if you use this to make a specific material look good it will also change how all the other materials will look that are in the lights path, you may fix one material but it will likely be at the expense of other materials, could end up just going around in a circle.

 

This PBR just uses the generic sun and a few recessed light fixtures. No Photoshopping other than adding my logo.

 

 

 

[Renerabbitt]

I am typically very satisfied with my PBR's, and I do absolutely no work on them aside from painting materials like anyone else would...difference being that my materials are custom and all have AO maps. 

I don't add lights

The only lighting I modify is the sun.

My scenes are usually done 2 minutes after I paint finishes.

This is of great value to me, good enough for my clients, if they want better, they pay for an external render:

 

[TheKitchenAbode]

Just a demonstration as to how impactful the sun can be. This scene has no active lights, everything is off.

 

Notice how drastically everything changes, colors, shadows, reflections, textures, visual sense of depth, background.

 

This is also an example concerning when to adjust materials. If you had the first scene displayed and attempted to correct it by playing with the materials you would likely end up with a big problem once you added the sun. You need proper lighting to see what the material really looks like, then and only then should you tweak the material.

 

Sun Off

 

Generic Sun Turned On, 25 Lux

[Designer1]

Wow more great examples Graham and Rene! 

 

Yes, the one texture that I really noticed was not looking realistic was the glass.  For the most part it just looked like an open space without glass.  Can I ask you Graham, you said you and Rene created a new texture?  How does that one behave in the PBRs much better then the standard glass options in Chief?  Do you remember the other post that you shared some of the examples of this new glass texture?

 

Thanks!

[Designer1]

Since everyone has been so kind to help me with the PBR stuff I wanted to share something I created yesterday that might help someone.  So I couldnt master Grahams lighting to lighten up the choppy shade of the PBR trees, so I found a solution that worked well for me.  

 

I got the multi copy tool and selected a 3d shrubs I liked.  Mixed them in a circle configuration.  Then used half of them as "reverse symbol" to vary the look.  I change the density of the shade by group selecting the bunch cad block it then just copy and paste the bunch together on top.  That way you can get the consistency of the shade your looking for ie: more or less sunlight.  Then select your cad block and raise it up to the height of a tree, above the camera and there it is the floating shade!

 

So obviously there are many configurations of plants, sizes and shapes but I like using these shrubs vs the 3d tree.  I hope this can help some of you.

[Renerabbitt]

3 hours ago, Designer1 said:

I got the multi copy tool and selected a 3d shrubs I liked.  Mixed them in a circle configuration.  Then used half of them as "reverse symbol" to vary the look.  I change the density of the shade by group selecting the bunch cad block it then just copy and paste the bunch together on top.  That way you can get the consistency of the shade your looking for ie: more or less sunlight.  Then select your cad block and raise it up to the height of a tree, above the camera and there it is the floating shade!

 

Would've saved you a ton of time using the Distributing tools in chief under Build/Distributed Objects/ ...just an FYI

My glass has a grey normals map, a trick discovered by Jintu I think, and is a general material with a medium roughness and transparency. I've modified a great deal of Chiefs out of the box lights, one by one in a set scene to make sure they play nice with each other for typical scenarios. As such, I don't often need to touch anything, set it and forget it. My template plan has all of the settings I want for PBR in a saved camera.

The second I need to spend more than 10 minutes setting up a PBR is the moment I decide I will render it elsewhere.