Glow me, C4D!

Man, I still can't get good results with frischluft. When I said it was working like a charm, I still had some artifacts around my edges, but I figured it was b/c I was using a depth map of the same resolution as my image (720x444). Well, I rendered a depth map 3x the size of that (2160x1332) and I'm still getting weird artifacts.

The parts out of focus do look beautiful... After inverting my depth map, I just compared the frischluft blur to AE's 'Lens Blur', and I agree that it does look a bit better. You can blur it more as well without the image becoming washed out like it does with 'Lens Blur'.

If I could solve these quibbles I would be inclined to purchase the plug-in, but idk at this point.

Here is a rendering of my image using frischluft w/ the depth map inverted and and downscaled from it's native resolution of 2160x1332 to 720x444...

frischluftDOFinverted.png
 
Some programs read the black and white in a depth map differently, so that's why the need to invert. In C4D, black is close and white is far, whereas it might be opposite for other renderers. You should be able to select the depth map, invert it, then dial in the "radius" setting as needed. Works for me, although I'm not using a double res depth map so I have some edge issues, but other than it's all good.
 
also, there is some weird glitching happening above the "u" and below the "c" in the dark "frischluft" text at the bottom...
 
Oh, so there is no universal standard for what polarity white and black equal in a luma matte... interesting.

Well I could live with that, but these glitches are a dealbreaker...
 
Figured it out, you need to blur the depth map by a few pixels, but due to rendering order you need to precompose it with the "Move all attributes into new composition" checked off. That way it will calculate the depth map post blur rather than pre blur as it would without being precomped.

fixed.jpg
 
wow... that looks great. I'll go try it.

Maybe a silly question, but why would you want to uncheck "Move all attributes into new composition"?
 
Do you get those same glitches if you turn your 'radius' up?

I just rendered a new pic at 1920x1080 with a depth map just to simplify things a little... and I'm still getting the weird edge artifacts and the glitches around the bottom logo... I put the depth map into a new comp, blurred it by a few pixels, then put that comp into my original comp and re-set my depth map to the new comp containing the slightly blurred depth map... and the artifacts are worse! Lol... I'm stumped.

I would just like the edge of the sphere to be in focus with the background out of focus. This is definitly proving to be quite the challenge...
 
I should add that the artifacts around the edge are MUCH less pronounced when using the 1920x1080 image. I guess I could live with it... but that would mean I would always need to use the effect with a high-res image.

I'm also thinking that... I'm still not doing something right somehow.
 
I think its because of the natural gradient that happens at the edge of the sphere in the depth map. Like most images, there isn't a straight fall-off from black to white when you look at the pixels on the edge up close. It wouldn't look smooth if there was no gradient. So... the same goes for the sphere in the actual image. Some of the edge pixels of the edge of the sphere are being blurred b/c the correlating depth map pixels are some shade of grey as they are transitioning from black to white. This causes the edge artifacts.

I don't know if there is a way to change the depth map image to minimize this... but, again, its pretty slight on a higher resolution image. Unfortunately, it's looking rather grim for a high-contrast area of depth on a low-res image...
 
I've also read around the forums that C4D's multipass' AA is different from the main AA render settings and that causes issues. It happens all the time with object mattes, causing you to have choke the edge by 1 pixel, and even then there's some strangeness. It could be part of your edge problem. Either way, just giving it a 1/2 to 1 pixel blur should help out, and, yes, things in the real world never falloff like so, it's always gradual, just a matter to what degree.
 
Cool, I'll check it out. I'm still studying the first one. The way you set up the mattes is what I'm trying to understand... I'm not familiar with doing this in AE. You have the 'multi-pass object (1-2)' on there twice, but invisible... and deleting those files still makes a difference to the composition. Hmmm...

Also, I just want to check and make sure I know what it is you are going for here. The goal of this project is to blur the refraction of the background while keeping the thing in the middle in focus, right?

The reason I ask is b/c I was also wondering how to completely seperate the sphere from the scene, but still interact w/ the rest of the scene with refraction, reflection, etc... I.e., if i deleted the sphere, I would want the image of the background behind the sphere to be complete... so the sphere would sit on top of a complete image.

Anyways, I could trick myself into thinking that maybe this is what you were going for with your project, but I think you are solely aiming at blurring the refraction in the sphere... But another thing I'm wondering is why you left out the sphere from the depth render...?
 
Sorry if I got confusing in my last few posts... I was solely sidetracked with and referring to tweaking in the frischluft plug-in to be able to get clean edges. I figured it out!

This probobly applies to any post-DOF effect as well... I needed to render my depth map with NO anti-aliasing at all. I picked this up from reading the Lenscare manual. It is like I was saying a couple posts ago, with anti-aliasing, the are between black and white has a gradient applied to the pixels to make the image look smoother. In a depth map, these pixels are varying shades of grey, feeding incorrect depth information to the DOF effect on the image.

For ultimate quality, one would render their original image at double their target resolution. For the image, rendering with 1/4 the amount of the desired anti-aliasing saves time b/c it will equal out once you half your resolution (if you want 4x anti-aliasing in your final image, render 1x at double resolution). Then, one would run a seperate render at the same resolution of just the depth map with NO anti-aliasing.

Taking these two images (the depth map and the image), one would then process the frischluft DOF in the usual fashion and tweak to their desire. Once finished, they then would either render to a new image or place the working composition into a new composition and reduce the resultion by half. This yields the best results.

I did this for the image we are working with and got the best results I have had yet. I wanted my final image to be 1920x1080, so I rendered and worked with 3840x2160 resolution images from C4D.

Here are my results (large image):

frischluftDOFHQJPEG.jpg




Now, the manual also says it would be acceptable to upscale an existing image to double resolution and repeat the same process... but I didn't try that so I'm not sure how the results would compare... I bet it would still look great though.

Anywho, now to get back on course with studying my multi-pass :).
 
Hey triple-

Thanks!

Basically, I just used a 'landscape' object, made it spherical, and tweaked the furrows. For the material, I used a texture of a rock and added some displacement mapping with noise as well as a bump channel with noise. The floor is a seperate landscape, not spherical, with the same material.

For the doorway, I just made the landscape editable and cut a hole in it by deleting some polygons. The light outside is a cube with dense white fog as a material. Right in front of this is an omni light with volumetric visible light.

The doorway and the material are a bit simple, but I did this on purpose knowing that I was going to be applying a DOF filter to the image. If I was planning on having it in focus, I would have dressed it up a bit more with a beveled border, thin plants hanging down, some loose rocks, etc... But, one key thing to modeling efficiently is having the foresite to know when such things are unnecessary due to how your final image is going to look.

The stelactites and stelagmites are simple cones of varying width and height. I just scattered them about randomly in the cave. Their materials are a bit more tricky. The bottom half is a white luminance w/ a bump. The top half is a copy of the same material as is on the cave, but tweaked a little bit with sub-polygon displacement. There is a gradient in the alpha channel of this material that allows for the blending of the textures.

Anyways, thats the jist of it.

This was part of a video I had begun to put together about a month ago. I have some other stills as well... I'll post a few of them.
 
These have all been scaled down from their original size... Also, I wanted to get them done by a deadline for a contest, so most of them are sort of works in progress...

energyonwater_00000.png


energypathways_00000.png


lookingon_00000.png


underwatergate_00000.png


woodshut_00000.png
 
Last edited:
Wow thanks for the response and really great work! That first picture with the water and broken doc was that all done in C4d? Everything Looks beautiful and I'm loving the water!
 
You must log in or register to reply here.
Back
Top