Most Liked Content

Deep Compositing : How it works in Fusion.
In the informative interviews done by Mike Seymour on FXGuide about deep compositing, he has described examples of fog and candle lights in a row as an explanation of deep composting. These can be achieved in Fusion with the Volume Fog Tool.

The volume Fog tool is a special case render that takes 3D camera and lighting along with ‘deep images’ and renders them out from the cameras point of view.
The ‘Deep Images’ have XY and Z so that it can fit a volume of a 3D scene. The WPP is used to occlude the depth, so beauty passes can be merged into the volume image.
So in Anonymous nonlinear fog volumes of varying density can be integrated with the 3D scene. A rendered beauty pass along with its WPP(RGBA-XYZ) is put into 3D environment of the Volume tool and reacts with the fog volume. Because of the World Position Pass makes a fixed reference in terms of its XYZ coordinates, a Fog volume will stay fixed correctly to the scene. Zbuffer on the other hand is from the cameras rendered point of view, so camera based depth cue fog is not fixed to the scenes point of view.

In Fusion, you can create volumes procedurally using tools like fast noise, or particles, you can also load them via EXR. By doing things procedurally or a mixture of loading and procedurally the amount of data to disk can be dramatically reduced.
In Fusion 6.3, adding lighting models to the Volume Fog tooling, takes what can be achieved in this new deep arena to a whole new level. GPU powering this development makes the Deep Volumes interactive/realtime when rendering.

thank You Steve for such in-depth explanation!
one learns everyday:)

Sorry… all this writing. Why not just create a KAVT (Kick Ass Video Tutorial)…? It is much faster than all of these technical descriptions.
Looking forward )
/Samsoe

How do I load a render list (and/or revise that list) before rendering actually starts? Seems like "Add Comp" just loads 'n goes, and I don't want it to "go" until I'm ready (especially if I don't want an existing previous render to be overwritten!).

Thanks,

--Kelly

the flipbook softwre jefecheck is now availible in version 1.5 and is no availible for free!

http://jefecheck.jefecorp.com/

So you have people in here mentioning "Kak" all the time, and how they love it. Sounds like your sense of humor alright, from what I remember Showed it to Jan E. earlier this week and he had to chuckle too.

Looking forward to give it a try and thanks for sharing.

Wow, it's so much better than the standard CornerPositioner!
Neat and logical interface.

But indeed, what I always missed in the CnP, was the possibility to drag multiple points at once.
And this tool still doesn't replace the Krokodove Positioner as it has no separate BG / FG inputs.

The advantage of Krokodove is that you can position corners of an image with different resolution over the background in the background's coordinates without preliminary reformatting.
Would be great to implement such functionality as the BG and FG image are often different resolutions and aspects.
So, if the FG isn't connected, the tool works just like it works now. If both image inputs are used, then it composites FG over BG and the points are transformed in the BG coordinates.
Perhaps, you could even borrow the Merge multibutton control from the Tracker to provide more output options (BG only / FG only / FG over BG (default) / BG over FG).

Concerning the Source / Destination choice, you may notice, that in the GridWarp when you switch to Source, it shows the original undistorted image which makes it easier to position the source points.
Although, you can always display the image from the previous node, so it's not so important...

In the standard CornerPositioner and in the Tracker the points are listed in the following order:
1. Upper-Left
2. Upper-Right
3. Bottom-Left
4. Bottom-Right
Do you also think it would be nice to stick to this standard?
I would even use those definitions (Upper-Left, Upper-Right, Bottom-Left, Bottom-Right) instead of the numbers to avoid confusion.

I hope, it doesn't sound as too much of critics, I wrote all this because I'm really looking forward to using this tool next time I need a corner positioning

Thank you once again!

Just spoke with Eyeon and they will transfer it all for $1,000 and upgrade to version 6x it for an additional $2,000 although the upgrade is not required.

Just to clarify, I am asking for $1,000 for 4 seats of DF 5x, 10 seats of RenderNode all on a floating license and HASP dongle. You will need to pay the transfer fee and optional upgrade costs if you elect to do that.

David Ellzey
Xpletive

Join our Facebook site at Pigs Fly Forums

ALSO please "like us" on Facebook at FPRstudios.

Thanks

Ray

I guess you are right. Although I feel like I should have chosen a less inflammatory title back then, it does reflect the emotional state many of us had when Fusion was practically nowhere to be found on related internet media.

A DI company has supplied 12 bit linear dpx files (the footage was originally 12 bit linear Log C prores from an Alexa) purely as a no loss wrapper format (not the most sensible of formats for them to choose given the lack of any lossless compression strategy but there it is) and I find that fusion can't read them.

This format is not that unusual - in fact I've seen 16 bit dpx files used as well so I tested that and fusion reads them fine but not 12bit. Nuke reads the 12 bit fine as do most other programmes I've tried including Flame (not Shake, though). I can convert them in Nuke and flame to 16b dpxs or tiffs or sgi or whatever for now (the prores 444 qts aren't available) but fusion should really be able to deal with this given the increase in 12 bit linear camera systems.

I'm using 6.4b 1104 64 bit. Anyone else hit the same problem? Is there a workaround within fusion? I can supply an example 12bit dpx frame if the fusion chaps need one to test. They might want to add 12 bit to the ChangeDepth node or maybe the dpx write node options as well as we are sometimes called on to supply files back in the same 12 bit linear format.

Hello Everyvody,

I'm new here and I'm sorry for my poor english.

I'm currently making a basic 3d scene with an image plane, add some bump, displacement, lighting and a camera move.
After that I just want to render it with the OGL renderer.
everything work perfectly and it render well on my machine.

The problem is that when I send it to the render farm, it doesn't work.

In the consol of backburner, I have an error message:
Renderer3D1: warning - lighting mode is set per-vertex (reason: shaders not supported)
Renderer3D1: Render aborted

The strange thing is that my lighting is set to per-pixel.

If I switchit to per-vertex, I have another message:
Renderer3D1: fatal error - failed to create GL context
Renderer3D1: Render aborted

Hi guys,
I'm just upgrade to Fusion 6.4 and I open some old comp from the last version, and all my depth blur node don't work, I don't have the Z channel.
Is that normal or a 6.4 bug, or do I have to set the channel booleen differently.

Thank you for all you answer.

I'm setting up a linear workflow for DPX files from a Quantel system. Though I feel this is a more general problem.

Basically LUT nodes and the 'Display View LUT' are producing different results with the same LUT files
The Display LUT seems to be clipping highlights (bad), while the various LUT nodes do not (good)

I use nvidea geForce gtx 285 as my graphics card. I'm not sure if this is the issue.

It is particularly bad using ARRI LogC. If you look at the A/B image bellow you can see that the sky in the A side is clipped while the B side is not.
The view of the first CineonLog1 node is clipped buy the Display view LUT. Even though the color Inspector still shows the colors to be super white (i.e. R,G, and B all have values above 1.0)



This is an issue because I have LUT files generated by the Quantel artist. He has designed a 'look' for the show and sent me LUT files to take me from Alexa DXP 'ARRI LogC' files to 'Rec709 with the Look'

For me to use a Linear workflow I need to find a way to go from 'Linear to 'ARRI LogC' and then 'ARRI LogC' to 'Rec709 with the Look' using a 'Display View LUT'

Fusion makes it simple to add two display LUTs but the clipping is a big problem for my workflow.
You can see in the image below that the 'Display View LUT' on the left is producing a clipped image compared to the right image using the LUT from the Quantel artist.


I've tried making my own LUT to using the LUT cube nodes. The clipping is still present.




Are other people having clipping issues with the view LUT ? Do I need to replace my graphics card, or is it a problem with the FUSE ?

http://www.anatomica...olor-reduction/


Ok, this might seem a bit weird, but this tool converts color to greyscale.

The neat part is that, unlike other methods in Fusion, this tool actually looks at the contents of your image to determine an optimal global fitting for the dimension reduction. You really have to compare it to something like ColorToMono or ColorSpace to get an appreciation of what it does.

Don't forget Krokodove's Switcher. Can switch between 32 different nodes and the connections are "wireless" so you don't have 32 pipes going to one big node...

What are available options for lens distort/undistort tools in fusion?

Thanks for posting and all the work on the AppLink!

If you do re-work it at some point with free assets, I'd certainly be interested in seeing it - and I'm sure others who are beginning with 3DCoat and Fusion would also!!

Hi,

another macro we are using often. It is a normal falloff and usable for either normal or position passes.
I think there was or is a tool in after effects, I always liked the idea to click on the normal pass and this indicates the lighting direction...
Hope it helps.

Johannes

SSAO Filter 1.00  ssao_filter_1.00.zip   19.34KB   156 downloads
Heya.

SSAO is a technique that was developed for the PC game Crysis in 2007. I’ve read papers published from Sigraph and made a Fusion plug-in that utilizes the technique the developers of Crysis invented. The technique is about how to extract a fake ambient occlusion out of a Z-depth pass.
The tool takes a specified amount of samples around each pixel, usually a 10 by 10 matrix and compares the depth with the current pixel to calculate occluding pixels. Furthermore I’ve implemented a depth to normals algorithm which is then used to calculate the SSAO.
The user can specify in the plug-in whether the SSAO should be calculated based on supplied normals (in NX-, NY- and NZ buffer) or calculated normals. This plug-in is currently under development and is highly technical and requires a lot of tweaking to get desired results.

Input a depth pass to the SSAO filter (stored in the Z channel) and it will return normals and screen space ambient occlusion in the RGB(normals) and A(SSAO) channels, the SSAO can be rendered based on either calculated or supplied normals, were supplied will give you best results. This is still a work in progress so don't expect production quality results.. : )

The plugin is multi-threaded and compiled for 32bit/x86.
Make sure you run atleast Fusion 5.3.55 with this one. : )



matrix: This sets the grid that is being sampled around each pixel.
amount: The amount of occlusion to add.
small depth: If the depth is less than this no occlusion is added.
depth threshold: The distance in depth that the filer looks in.
normals angle control: Controls what angle diffrence that should be allowed, in order to avoid occlusion on flat surfaces.
normals z control: Boosts the Z calculation for normals


Good luck!
Christoffer Hulusjö
christoffer (at) hulusjo.se
http://christoffer.hulusjo.se

Give it a try and let me know if you have any suggestions or feedback.