Week 1 – Nuke 3D:
What was covered in the lecture:
3D Side of nuke:
– TAB to view the 3D viewer.
– Same controls as Maya
– There’s an option to remove the grid
– 3D nodes are curved
To assign a texture to an object:
The Workshop Activity:
1. Space Camera:
Go to 3D viewer. Bring in a cylinder and a checkerboard. Bring in a cube and a colorbars. Link them. In order to view both of them add a scene node. 2 3
Bring in a camera. To see through the camera change the view from different to the name of the camera. Link the camera to the scene.
To see everything back to 2d environment add “ScanlineRender1” node and link cam to camera and obj to scene. 8 9. To animate the camera set a keyframe and move the camera.
We can change the motion blur by adjusting the samples under ScanlineRender node MultiSample Tab.
2. Tunnel Cam Projection
We already have animated camera. Then bring in a tunnel sequence.
Then bring in a cylinder and project the tunnel on it. Set the cylinder height to 100. Rotate it.
Add Project3D1 node between them. Add ScanlineRender2 and link like so:11
Method 1:
Then duplicate the camera, go to the first frame and delete the animation of it. Link it like so. The first camera is for the still image and the second camera is for the movement.
Method 2:
Add a frame hold on the first frame.Then link everything like so:
Final Result:
3. Prison Cell
Bring in all the images and roto each part:
Add project3d node and link it to the projection camera. Then add 3D card for each of them. Place them correctly in the 3d environment.
After each card link it all to a scene node. Under the scene bring in scanline render node.
After we are satisfied with the scene, delete the scene node. Instead link to each of them “ScanlineRender2” node. Link the camera to the animated camera (green).
Add a merge node to each of them so we could adjust other nodes individually in the future (like defocus).
Week 2 – Nuke 3D Tracking:
What was covered in the lecture:
Key Concepts
– The things closest to you moving fastest, the more far they are the slower they objects will move
– There are two steps: Tracking and Solving
– Take into account lens distortion. Using checkerboard to fix the perspective.
3D Clean up Script layout and VFX Pipeline for integrating 3D Elements:
The Workshop Activity
3D Tracking Cleanup:
On timeline:
I = in | O = out
Set the timeline to in/out
On CameraTracker node:
To view the tracking marks:
Delete the water using roto and copy nodes (because it’s moving and it will cause some problems to the tracking).
To mask out our roto, under camera tracking change the Mask to “Source Alpha”. Change the range input if needed.
Hit track. Then hit solve. Go to AutoTracks tab, select Delete Unsolved (orange) and Delete Rejected (red). Update Solve.
Export as a scene (because we will insert 3d elements afterwards).
See through Camera1. Always switch between default to camera1.
Wipe of original footage after UD and the scene like so. Bring in a card, link it to the scene and disable it. Select Choose vertices and select 2 vertices. Go back to node selection. Enable the card and choose match selection, position.
Go to the top view and match the location of the card to the window.
Bring in ScanlineRender node to convert the scene from 3D to back to 2D. For the background pipe link a reformat.
Add a merge node. Link A pipe to RD and B to the original footage.
Create a Patch:
Final Script:
Week 3 – 3D Equalizer:
What was covered in the lecture:
Shortcuts:
Z – Jump to start frame
X – Jump to end frame
A – Jump to previous keyframe
S – Jump to next keyframe
D – Delete keyframe
V – Step one frame forward
B – Step one frame back
Q – Track one frame forward
W – Track one frame back
E – End/Start point
R – Flip track direction
T – Track
Y – Center 2D
F – Full frame view
G – Gauge Marker
F1 – Overview Controls
F2 – Manual Tracking Controls
F3 – Distortion Grid Controls
F4 – Autotracking Controls
F5 – Lineup Controls
F6 – 3D Orientation Controls
Alt+1 – Object Browser
Alt+2 – Deviation Browser
Alt+3 – Timeline Editor
Alt+4 – Curve Editor
Alt+5 – Attribute Editor
Alt+6 – Parameter Adjustment Editor
Alt+7 – Image Controls Window
Alt+8 – Python Console Window
Alt+9 – Online Help Window
Alt+0 – New 3DE Window
Hit space to play/stop
How the software look like:
Open the software, change the view to basic and close the following windows:
To set up shortcuts: right click, set shortcut, aquire, hit the key, and use.
The Workshop Activity:
Overview Exercise:
Double click and select our sequence.
Change to 25 Frames per second:
Double click on lenses and rename. Change film width, focal length and pixel aspect
Export buffer compression file:
To set a tracking mark CMD+Click
On frame one create a tracking mark. Hit G and then T to track
Option+click to deselect the marker To delete point – hit delete button
Option + C – Calc all from scratch = solve – After setting a tracking marker
F6 = 3D Viewer. F5 = Lineup Control
How to track:
To track forward: On frame one create a tracking mark. Hit G and then T to track
To track backwards: On the last frame create a tracking mark. Hit G and then T to track
To track from the middle: Track until the end. Then hit R (Flip track direction) and E to change from start frame to key frame. Then G and T.
To track a marker that disappears at the middle: Track, then go o back to frame where it is still visible, adjust the track and hit W (write), do it couple of time, when it almost disappears. Then hit E (end).
To Visualise the graph with the “errors”:
Config -> Deviation Browser
Deviation Browser -> Show Point Curves -> All Points
To reduce the error:
Go to parameter adjustments (option + 6). Keep only the focal length.
Then calculate again (option + C). The error will be reduced.
Fix lens distortion:
Go to att editor and change “lens distortion model” to “3DE Classic LD Model”.
Go to parameter adjustments (option + 6). Remove the focal length. press “clear” then “Adjust” then calculate again. (option + c)
Arrange the 3D env so it will be above the ground:
Alt+shift to select multiple points. Select 3 points that are on the floor.
Go back to 3D env. Edit -> Align 3 points to -> XZ Plane.
Then go back, to 2D, select one point, go back to 3D environment, edit -> move one point -> origin
Adding a 3D Cube:
First of all, select all the points. Geo -> Create -> Locators.
Select a point (option+drag), go to 3d viewer (f6), 3d models -> create -> cube. Select the cube as well (shift+option+drag). 3D model -> snap to point.
Scale it down and rotate
Week 4 – Cameras and Lenses
What was covered in the lecture:
Different sensors:
How Does Anamorphic Works?
Video (red) vs Full Frame (green):
The Workshop Activity
Nodal Sequence:
Nodal sequence means fixed position of the camera but its panning
Browse the sequence and change the frame rate to 25. Then go to Import and choose “import buffer compression file”.
Now, we need to add to the software the lens information.
Double click under lenses.
Change the name of the identifier. Change the focal length, film width, pixel aspect, and to make sure the film aspect stays the same
Canon 5d mk3 35mm:
Filmback width: 3.6
Focal length is the sensor = 35mm
On frame 78 create a tracker (ctrl+click). The G to gage it. Center 2D. T to track. Deselect Alt+click. And the track again. Create 10 trackers. 6 of them should be in the scene the entire time. Then ALT+C to calculate it. Then open the deviation browser. Then under deviation select show all points.
Under Camera Constrains change to fixed.
To show the graph on top: Go to Config -> Deviation browser. Deviation Browser -> All Points
Under camera settings, change the focal length to adjust and make sure its on 35 (on the pic its 43 – ignore that)
Hit ALT+6 to open “Parameter Adjustments” window and keep only the focal length. Change it to adaptive and click Adjust. Then Alt+C to calculate again
Go back to camera settings. Under Lens Distortion Model change to “3DE Classic LD Model”. Also. enable “Distortion” and “Quartic Distortion”
Alt+6 to open this window. Hit “Clear” and then remove “Focal Length”. Select adaptive all, and then “Adjust”. Calculate again (Alt+C)
Go to 3D Orientation View (F6).
Select the camera (Alt + Drag). Make sure you select just the camera and not the scene as well. Because we have the information of the height of the camera from the floor we can adjust it to be even more accurate. At the bottom click on “Local” (it will then change to “Global”) and change to 170.
Week 5 – Free Camera Movement Tracking
The Workshop Activity:
Free Move Camera:
Open the sequence and change the frame rate to 25.
Go to Playback -> “Import Buffer Compression File” to load the file.
Change the name of the identifier. Change the focal length, film width, pixel aspect, and to make sure the film aspect stays the same
Rename to: Canon 5D mk3 35mm
Filmback width: 3.6
Focal length is the sensor = 35mm
At the end change the “Pixel Aspect” back to 1.
Go to frame 4, as it’s sharp. CTRL+click to create a tracking point. G to gage, T to track.
At some frame it will stop tracking. Adjust the tracking point and the searching box, hit G then T. Repeat the process until the end of the sequence.
To jump between keyframes the shortcut is A and S.
To track backwards from frame 4: Go back to frame 4. Hit R (Flip track direction) and E to change from start frame to key frame. Then G and T.
To track one frame at a time: Q/W
Go to frame 151. create a tracking point. G then T. Track one frame at a time (Q/W).
When there’s a lot of motion blur just split the difference and put it in the middle
After we finish tracking forward, track backwards from frame 151.So go back to frame 151. Hit R then E. Then G and T. End the tracking point at frame 44 by hitting E.
To track the bottom step, make sure when you select the corner, the search box shouldn’t cover the background as the background will change during the sequence and it won’t track successfully
Repeat the process and create around 40 points.
Second Part:
Open the completed scene which includes 31 points. Import the same sequence again and set the frame rate to 25.
To Visualise the graph with the “errors”:
Config -> Add horizontal pane -> Deviation Browser
Deviation Browser -> Show Point Deviation Curves -> All Points
To select a specific curve: Alt+drag
To turn off the image plane: View -> Show Imageplane -> None
To reduce the error:
Go to the lenses and change the focal length from fixed to adjust.
Once it’s changed to adjust it will appear in our Parameter adjustments window (ALT+6)
Make sure its on “Brute Force” – exactly like in the picture. Hit adjust. Then calculate (solve) Alt+C.
Open again the attribute editor (double click on lenses). Enable Distortion and Quartic Distortion. Then close the window. Open Parameter adjustment window. Remove the focal length. Hit “clear”, change both of the to adaptive then hit adjust.
Then calculate again (Alt+C).
To scale the scene to the actual dimensions:
Select these points in the lineup view (F5). Then go back to “3D orientation controls” view (F6). Constraints -> Create Distance Constraint. Add in the distance between these two points.
To align the scene to the grid:
Select 3 points on the floor (Alt+Shft+Drag). Then go to 3D view and Edit -> Align 3 Points To -> XZ Plane
To set the floor:
Select a point we want to set as our origin. Edit -> Move 1 point to -> Origin
Create Locators:
Under Geo Objects select all of the locators. Geo ->Create -> Locators
To scale them down: Geo -> Scale Down Locators
Create a Cube:
3D Models -> Create -> Cube. Then select a point on the floor and the cube. 3D Models -> Snap To Point
Rotate and scale it down:
Week 6 – 3DE Freeflow and Nuke:
What was covered in the lecture:
In this week we covered how to bake a camera on 3DE, export cameras, locators, geometry and distortion.
We learnt how to import it to nuke and also how to install plugins.
Then we covered how to create and apply patches.
What was covered in the lecture:
Export to Nuke
Open 3DE:
Open the project, then import the sequence the same way when you open a new file.
Change the frame rate to 25.
Bake:
First we need to lock our camera = bake it
Go to 3D orientation. Edit -> Bake Scene
Then 3DE -> Save As and then use the same version name but add “Baked”
Export Camera:
3DE -> Export Project -> Nuke
On start frame write 1001 (same as our sequence) and select a location. Save it under cameras and name it Cam_v001
Export Locators
Go to Point Group -> Geo Objects and select all the locators
Then go to Geo -> Export OBJ
Save it under Matchmoving file -> Geo. Name it locators_v001
Export Geo
Go to Point Group -> 3D Models and then select the cylinder
3D Models -> Export 3D Models as OBJ
Save it under Matchmoving file -> Geo. Name it cylinder_001
Export lens distortion
Go to File -> Export -> Export Nuke LD_3DE4 Lens Distortion
Save it under Matchmoving file -> Undistort. Name it LD_v001
Nuke
Bring in the image sequence. Change the Framerate to 25 and the size format
Bring in the camera, geo and locators.
For the camera, delete everything but “FreeMove_v0011” 
Install a distortion plugin:
Put the plugins inside .nuke folder. Then update it.
Un/Distortion Pipeline:
Undistort:
Add “Blackoutside” node, then a copy of our distortion node. Underneath link a reformat node. Change the Type to Scale, Scale to 1.01 and resize type to “None”
Redistort:
Bring in a reformat node and set the “resize type” to “none”. Make a copy of our distortion node and change the “direction” to “distort”
GENERAL PIPLINE
Bring in a “Scene” node. Link to it our locators, cylinder and freemove camera.
F to frame the view of the camera.Change the camera from default to “freemove” to see through our tracked camera
Bring in a wireframe node and link it to the cylinder, change the settings to “see through” Bring in a constant node and link it to the locators so they will be visible. Link the scene node and the camera node to a “ScanlineRender” node
We can see now that our locators are not exactly were they should be. To solve this:
Under our freemove cam, under projection settings, duplicate the horizontal and vertical aperture by 1.1 (same number we set on our reformat node).
Copy the two nodes under Redistort backdrop: reformat – resize type = none and “LD_3DE4_Freemove_v2” direction = distort. Link everything like so:
Create a patch for the sign
3D Patch:
Add project 3D node and link it to our scaled camera. Link a framehold between them.
Link a card below it and then “ScanlineRender” node. Link the Scanline Render like so: obj to card, cam to scaled camera and bg to reformat overscan.
Go to 3D view and place the card wherever it should be. Change uniform scale to 100. Make sure to rotate it as well viewing the top view.
At the bottom, link our Redistort nodes.
To add motion blur, go to ScanlineRender node, under “MultiSample” tab, change the samples number to 4.
To visualise the tracking markers and the cylinder, add merge node above the redistort and link everything like so.
Final Script and Video:
Week 8 – Surveys
What was covered in the lecture:
In this week we covered how to bake a camera on 3DE, export cameras, locators,
The Workshop Activity
Survyes:
Bring in the sequence and change the lens settings:
Filmback width: 3.2
Pixel Aspect: 1
Rename the identifier: Canon 5D mk3
Change the regular units from cm to m for an easier set up: Option -> preferences
Under point group – right click->add new->point
Set the angle of 3d viewport according to the survey data (xyz):
Double click on the point and change to “Exact Survey”
Add one more point. repeat the process and start to build up the scene: 
Based on the survey data build up the carpet.
From the far cornet of the carpet, duplicate that point (edit -> duplicate survey point). Then double click on the point and change the measurements.
After creating the scene. select the origin point and F2 to go to our 2D space.
Go to frame 4 (our hero frame). Drag to create a box at the edge of the carpet. We won’t press ctrl because we do not need a new point.
Select the 4th point. because the area is quite dark, we can bring in colour control by hitting Alt+7.
Track these 5 points. Then calculate. (ALT+C). Then add 2 new trackers on this wall. Convert them to Surveyed (F6 – Edit ->convert to surveyed). Then calculate again.
2nd exercise:
New project. 3DE -> File -> Import -> import multiple frames
Just select the folder. Do not select the sequence. Press okay.
Select the second one and delete it.
Right click on the first image and select set current camera. To view these images hit v-b. Rename the Lens to Ref Lens.
Bring in the image sequence and change the frame rate to 25. Create one more lens by right clicking on the lens -> add new -> lens. Then, assign this lens to the new sequence we brought in by (Right click -> Modify -> Lens -> lens 2)
Adjust the settings of the ref lens and lens 2 like so:
Now, Set Current Camera on the first image reference. Set a key (ctrl+click). Then G. Then move to the next reference images and if this point is visible there as well, just click and gauge (without ctrl). We need to do between 30-60 points so it will work properly.
To share these points with the image sequence:
Set current camera to the sequence. Select a point and just click (without ctrl) where it should be and track it.
Week 9 – Shooting
What was covered in the lecture:
This week in class, we filmed outside university for our second assessment project.
We placed tracking markers, shot the scene, and took measurements for our survey data.
We also captured a 360-degree shot to record the lighting environment.
It was a fun and valuable experience, and after that I felt confident about repeating the process on my own at other locations.
Week 10 – Nuke Maya Pipeline
What was covered in the lecture:
In this week we covered how export to Maya from nuke, adjust in Maya and then import back to nuke.
The Workshop Activity
Export the project in 3DE:
Export the lens distortion info.
Then export the project like so:
3DE -> Export Project -> Maya
change it to 1001.
In Nuke:
We need to put overscan value to prepare it to maya.
Bring in the image sequence. Set the frame rate to 25 fps.
Add BlackOutside node, UD node and reformat node.
Inside Reformat node change to scale and 1.1. Also change it to none.
In Maya:
Set up a project. Change to 25 fps. Open a scene -> Mel Script.
Save the camera only as FBX.
To view through the camera go to panels and choose the camera. Set the image plane to our UD plate that we exported from Nuke.
Now because we have done the overscan, the tracking markers are not in place.
Under the camera settings, under Filmback -> Camera Aperture add before the parameter “=” and after *1.1 for both of them. The colour will change into purple.
Under image plane shape -> Placement -> Size: Add before the parameter “=” and after *1.1 for both of them.
Bring in lights and change the AOV Light Group name.
On the render settings, under AOV’s tab add RGBA. Inside RGBA tick “All Light Groups”
Change the image size in the render settings to the same size shown in Nuke.
In Nuke:
Create undistort script (reformat = none).
Create a script like so:
Week 11 – Presenting our work
What was covered in the lecture:
This week in class, we had the opportunity to present our work for the assessment and receive feedback.
I presented my medieval castle project and received the following feedback:
– The flag should be placed above the door but not aligned in a straight line, and it should be much larger.
– The geometry of the stone blocks should be adjusted using the lattice tool in Maya to make them look less like typical CG elements.
– Add pebbles to the floor to avoid it looking too clean.
– Consider adding torches to the left wall (I’ll need to test how this looks visually).
– Place another statue on the left side of the wall, rather than directly across from the right statue, to create asymmetry.#
After the presentations, we were given time to continue developing our projects and track shots if needed.
Week 12 – Josh Parks Masterclass
What was covered in the lecture:
In this week’s class we were given the time to work on our 2nd project and receive feedback from Angus and James. Then, on the last hour of the session we had a masterclass with Josh.
I presented both my Castle and Ant projects with focus on the castle one.
This is the shot I presented:
These are the things I thought I have left:
– Blend the CG better so it won’t seem like CG
– Adding a fog so the scene will have “separate layers”
– Add motion blur to the CG and the patches I made (I rendered a motion vector pass for the CG too)
– Add light coming out of the small room
– Maybe add a CG animated flag
– ZDefocus
The feedback I received was to change the colour of the background to match the scene, add animated elements to the scene so it wont be static and add a point of interest.
First Assessment:
Process:
1. Minimum 40 points tracked:
I set the attribute editor of the project and the camera according to the provided information.
Identifier: Canon 7D 24mm
Filmback Width: 2.23
Focal Length: 24
Pixel Aspect: 1
I tracked 44 points. After the first solve there deviation was 0.4382.
2. Get a low deviation and therefore a solid track
Focal Length:
Under attribute editor I changed the focal length from fixed to adjust. Then opened the parameter adjustments window (ALT+6), kept only the focal length and changed to adaptive. Adjust and then calculate again.
Deviation = 0.4144
Lens Distortion:
On camera settings, under Lens Distortion Model I changed it to “3DE Classic LD Model” and enabled “Distortion” and “Quartic Distortion”. Then opened the parameter adjustments window (ALT+6). Hit “Clear” and then removed “Focal Length”. I changed both of them to adaptive. Then clicked on “Adjust” .
Then I calculated again and the deviation was reduced.
Deviation = 0.3986
3. Solve and align scene in 3DE using survey data
Scale the scene to the actual dimensions:
The counter’s length is 1.4m
So I selected to points on the edge of each side of it in the lineup view (F5). Then on the “3D orientation controls” view (F6): Constraints -> Create Distance Constraint.
I entered 140cm (1.4m = 140cm)
Align the scene to the grid:
I selected 3 points on the floor. I selected them taking into consideration their deviation, visibility along the scene and how much floor they cover.
Then on the 3D view: Edit -> Align 3 Points To -> XZ Plane.
Set the floor:
I selected a point in on the floor around the middle on the sequence that I wanted to set as the origin. Edit -> Move 1 point to -> Origin
Result of: solve and align scene in 3DE using survey data
4. Add in locators and 3D Cube to test stability of solve:
Create Locators:
I selected all the points. Then, Geo -> Create -> Locators.
If I ever needed too scale them down: Geo -> Scale Down Locators
Create a Cube:
On the 3D viewport, under 3D Models -> Create -> Cube. Then, I selected a point on the floor and the cube. 3D Models -> Snap To Point. I also added one more cube and snapped it to one of the points located on the sign I will later in nuke.
4. Export camera, LD data, locator geo, and 3d models to Nuke
First thing first, baked the camera and saved a baked version
Export Camera: 3DE -> Export Project -> Nuke
Export Locators: Under Geo Objects, select all the locators. Geo -> Export OBJ
Export Geo: Under Point Group -> 3D Models, I selected the cube. 3D Models -> Export 3D Models as OBJ
Export Lens Distortion: File -> Export -> Export Nuke LD_3DE4 Lens Distortion
I saved all the information above on these folders:
5. Undistort lens and reformat
Installed 3DE plugins to Mac:
I added 3DE plugins for MacOS (osx_arm64) to the following folder:
Distortion:
Undistort:
Added “Blackoutside” 3de, distortion node and reformat node. Changed the Type to Scale, Scaled to 1.01 and changed the resize type to “None”
Redistort:
Added a reformat node and set the “resize type” to “none”, linked 3de distortion node and changed the “direction” to “distort”
6. Place cleanup patch on the fire exit sign
Created a regular patch:
Created a patch on the last frame (1111) and added a grade node to match the colour of the patch throughout the scene. I linked it right below the undistort process. Under the patch I created a 3d card so I would project the patch on it
I placed the 3d card in the
7. Overlay locators and 3d model to highlight strength to track
I linked to a scene node 3de locators, camera and cube that I have exported earlier.
I scaled the camera up to deal with the distortion and format of the sequence (under projection settings, duplicate the horizontal and vertical aperture by 1.1- same number I set on the reformat node).
Then linked everything to Scanline render node to convert it back to 2D environment.
Final Result::
Final Script:
Final Video:
Without markers:
2nd Assessment:
Process:
Brainstorm Summary:
I wanted to create a shot where I could practice, learn, and improve my compositing skills. My goal was to integrate a CG environment into a live-action shot, something I haven’t done before.
Here are some of the images that inspired the shot I’m about to explain:
The Idea:
I chose a specific location at the Tower of London, which I found using Google Street View.
I wanted to adjust the live footage so it would look like it was taken during the Middle Ages, showing a medieval castle instead of a modern tourist site.
My plan is to remove the bin, replace the metal railings with stone ones (which I’ll model in Maya), remove the modern building in the background, replace the floor with stone, and make a few more adjustments to complete the look.
Reference Image:
I want to match the colours of this scene to the scene i’m creating
Filming:
I used Blackmagic pocket camera with a 24mm focal length.
I filmed here:
Here is the first frame of the plate:
Measurements:
3D Tracking:
I set the attribute editor of the project and the camera like so:
Identifier: BlackMagic Pocket 24mm
Filmback Width: 1.896
Focal Length: 24
Pixel Aspect: 1
Get a low deviation and therefore a solid track
Focal Length:
Under attribute editor I changed the focal length from fixed to adjust. Then opened the parameter adjustments window (ALT+6), kept only the focal length and changed to brute force. Adjust and then calculate again.
Lens Distortion:
On camera settings, under Lens Distortion Model I changed it to “3DE Classic LD Model” and enabled “Distortion” and “Quartic Distortion”. Then opened the parameter adjustments window (ALT+6). Hit “Clear” and then removed “Focal Length”. I changed both of them to adaptive. Then clicked on “Adjust” .
Then I calculated again and the deviation was reduced.
Deviation = 0.3572
Solve and align scene in 3DE using survey data
Scale the scene to the actual dimensions: 1.87m
So I selected to points on the edge of each side of it in the lineup view (F5). Then on the “3D orientation controls” view (F6): Constraints -> Create Distance Constraint.
I entered 187cm (1.87m = 187cm)
Align the scene to the grid:
I selected 3 points on the floor. I selected them taking into consideration their deviation, visibility along the scene and how much floor they cover.
Then on the 3D view: Edit -> Align 3 Points To -> XZ Plane.
Set the floor:
I selected a point in on the floor around the middle on the sequence that I wanted to set as the origin. Edit -> Move 1 point to -> Origin
Result of: solve and align scene in 3DE using survey data
Add in locators and 3D Cube to test stability of solve:
Create Locators:
I selected all the points (53). Then, Geo -> Create -> Locators.
Then I scaled them down: Geo -> Scale Down Locators
Bake the Camera:
First we need to lock our camera = bake it
Go to 3D orientation. Edit -> Bake Scene
Then 3DE -> Save As and then use the same version name but add “Baked”
Export camera, LD data, locator geo, and 3d models to Nuke
Export Camera to Nuke: 3DE -> Export Project -> Nuke
Export Camera to Maya: 3DE -> Export Project -> Maya. change it to 1001.
Export Locators: Under Geo Objects, select all the locators. Geo -> Export OBJ
Export Geo: Under Point Group -> 3D Models, I selected the cube. 3D Models -> Export 3D Models as OBJ
Export Lens Distortion: File -> Export -> Export Nuke LD_3DE4 Lens Distortion
In Nuke:
We need to put overscan value to prepare it to Maya.
I brought in the image sequence and set the frame rate to 25 fps.
Add BlackOutside node, UD node and reformat node.
Inside Reformat node change to scale and 1.1. Also change it to none.
Versions in 3DE:
Castle30_V5 – Before lens distortion
Castle30_V6 – After Lens Distortion, Focal Length, Alignment and Survey Data.
Castle30_V7 – Added Locators
Castle30_V8 – Baked
Modeling and Texturing:
While modelling, I tried to make the left wall not perfect to make it look less like a CG.
I aligned the walls to the tracking markers.
Rendering:
I have 3 different lights:
General Light – Sky dome’ left wall light and right wall light
I plan to render different layers so it will be easier for me to comp it after:
– Shadow pass
– CG floor wall
– CG Statue
– CG torch
– CG left wall shadow mat
– CG flag
After bringing back the distortion
Compositing:
Sky Replacement:
I used the same plate, and because the plate was shot with raw files I could get the sky detail back.
Also, I removed the modern building in the background 
Cleanup:
I removed from the plate all the modern elements.
I used Mocha Pro to get a tracking information.
Plate Colour:
What I have so far:
What I plan on doing:
– I’m trying to make the scene look like this reference.
