Lighting

The lighting session this week was about getting the car in to the maya scene and trying out some renders with it, as well as use the AOVs that KK fixed in my scene. So we began by importing the latest version of the lotus model and place it in to the maya scene, try out a render and we noticed that the geometry of the car was in low poly. KK tried to figure it out but then I tried to press one, as I had done some modeling on maya before, when pressing 1 the geometry converts in to high poly.

After that was fixed KK asked me to get the ground color ready for next meeting as this was a shorter one, due to the amount of work from this final weeks. KK asked me to see if I could try to render something on the render farm from maya as well as ask Luke or Christos if it was possible to render from nuke on the render farm.

Matchmove

I finally got more points on the CP_0400 shot and booked a session with Dom for next week, as I tried to set the meeting to late in the week, so Dom was already booked. But I think I got better results this time.

This week I had 2 meetings with KK and Giulia and one meeting with Dom, I had issues with the completing of the lighting tasks, KK asked us to match the lighting, prepare the macbeth chart and try to set the AOVs and render layers. The AOVs did not work for me, not the disabling of the image plane and skydomes.

Lighting

We began by rendering the reference spheres, which I had issues with because I could not disable the image plane nor the skydomes from the render so the result was not the expect, but KK showed us a way to set up a “button” on a new shelf in maya where we could turn off and on the image plane.

The skydomes are deactivated by the render layers which is also something very confusing to me, but we figured it out and were able to rende the spheres to get to nuke, on giulias’ machine. I kept having issues so KK fixed them during the week and got back to me on this weeks’ second session.

This now has to be imported to nuke to get the light matching.

With the spheres in nuke we can create another viewer window and match the lighting with the footage through a bounding box made on the lighter area of the grey sphere and the same process for the chrome sphere.

After this, KK showed us a cool way to solve this hdri problem, by croping the chrome sphere from the footage and with a spherical transform node get the “HDRI” displayed in it.

Ofcourse, the result was not the best but we got the idea of how to see what really is on the chrome sphere. On the second meeting, KK fixed my scene and I was able to export the spheres correctly, and he taught us how to make a projection on the geometry using the node editor in maya.

Projection from viewport:

Projection from render cam:

After this KK showed us how to get the Macbeth Chart ready to import to the scene and have it as light reference. On nuke, we must rotate the image until we get it aligned, crop it and use the schew setting on y and x and distort the image until it is as much as possible like a square.

While treating the macbeth chart we noticed the footage had already been edited to look like a night shot so we got the values closer to the original before exporting the final chart image.

• 

• 

Matchmove:

On matchmove this week, I had a meeting where he told me the best way to get the points to track on a blurry and dark image was to use image controls on black and white or with the saturation at its maximum. I had already set a few points, that Dom said were good and now I must continue to make sure I can finish every task I signed up to.

This week I focused more on lighting, although I did begin to track a shot, and set a meeting with Dom for next week (has we was already booked for this week) to get some tips on how to track such blurry shots. Also I had 3 meetings with KK and Giulia ( a student from the VFX MA) joined the team, which was great as help in these sorts of tasks is never too much. As she also knows a bit more about nuke, I was super glad to have help.

Lighting

This weeks’ task was to prepare the maya folder with everything on it, as well as create folders inside the scene and render folders for me and Giulia, we divided the shots for the 2 of us. I got the reference footage for the z004 and Giulia got the z002, these shots are not going to be used in the actual movie but will be usefull for lighting purposes. The woman carries 2 spheres and the Macbeth chart which we will have to prepare and get the right values.

z002:

z004:

KK had asked me to prepare a maya scene and get the spheres in the right place, with some HDRI we found. This is the one we are using as it is the one truer to the footage.

After setting up the scene and placed roughly a camera on the right place we created the road geo and added a character, I chose Amanda rig, as I really like this rig and got the right materials and attributes to the spheres, this is how ir tlooked like:

At first, I was a bit confused why we needed to have the skydomes if we were going to have the cameras from matchmove and I believed that that would be it, but later I understood the whole purpose of this.

The lighting task for next week is to import a render of the reference spheres alone and get them to nuke to match the lighting as well as the macbeth chart.

Matchmove

This week on matchmove, I was able to track a few more points on my CP_0400 but as it was taking too much time, I honestly focused more on getting the lighting tasks done, as there were more people on the matchmove role than lighting.

This week I had 2 sessions with KK as the footage arrived and as it was delayed we needed to get moving on the lighting task.

Lighting

In the previous session KK asked me to make the nuke set up for the hdri we had found to get the high value and low value, I had an issue when finding the high value as the node has 2 outputs and on the viewer panel I wasn’t able to see anything. KK showed me a way to discover if there really was no information on that node, by hitting ctrl + mouse click on the viewer we can see if the values of the rgb channel change. They did change, so, the way to fix this was to change the outputs of the nodes and the problem was solved.

So the clamp node sets every value above one to one and the merge node (minus) is the hdri minus the clamp node info which will help us discover the location of the light source.

HDRI:

• 

• 

Clamp node:

• 

• 

Merge node (minus):

• 

• 

As we can see the hdri contains all the info, while the clamp node clamps that info and the merge node is all about the light source and intensity.

Before exporting values, we tried to debug the skydome issue from the previous session and discovered that the path for the hdri was different and tried a few more renders with the spheres and another issue appeared: the floor was not showing up, so KK was going to look into it and let me know next session.

In the middle of the week, the footage came! I sent it to KK and we had a session on friday to see what needs to be done. The first step was to organize everything on nuke and separate the shots for the different light rigs that we need to make. Every shot we had access to was imported to nuke and in the end we got a small video with the videos referenced with labels and blacking out when finished.

We uploaded this to ftrack and KK asked me to find HDRI matching the light from the footage, prepare a maya folder with everything needed inside, to set up the spheres in a maya scene and to see if I was able to track a shot (that turned out to not be necessary for next session).

Matchmove

When the footage arrived, I set a meeting with KK and while I waited I began tracking a shot, which was challenging as the footage is a bit blurry as it moves too fast, but I think this is a good practice, I don’t believe the footage in industry is always perfect to track like the shots we began working on. Eventually I stopped working on matchmove and prioritised the lighting tasks as it is much more complex than I thought.

This week I tried to focus more on lighting as I am the only one doing it and really need to understand it better. I had one more session with KK and this one was focused on the render layers on Maya, I had used it before but with little success so KK asked me to prepare a maya scene with some render layers and we would check it out and talk about that subject more deeply.

Lighting

On the maya scene I created a floor, a grey sphere, a chrome sphere and a skydome, assigned the materials to the spheres and setup the render layers.

The rest of the layers contained the same setup, so we began by checking if everything was correct then, KK drove the session on my pc through teams, also something that I’d never done but it is pretty handy as someone can show you everything on your own computer.

My render layers were okay, but we made a lot of changes as everytime we think about a light or a shadow that we want another render layer is needed so it can get out of hand pretty quickly so, one of the main focuses was the “housekeeping” to keep everything organized and intuitive. KK showed me how to ad content from outliner to the layers through expressions, a simple * in front of an objects’ name, this is usefull for when we are dealing with a lot of content.

When rendering the spheres we found a few issues with the hdri as it was not being disabled by the render layer so both of them were in display. For some reason Maya was using the master layer when we refreshed the render. KK is still going to look into the issue with the render layers as they were not working the way we want it to.

Matchmove

This week on matchmove I had an issue with tracking a second shot from the preparation videos as I tracked it but the points were not creating a depth on the footage, this was due to the fact that the camera that recorded that footage did not move enough for the software to recognise a paralax.

On my session with Dom he showed me a way to solve this, by creating two planes perpendicular to each other positioned in the limits seen in the footage (the end of the road, begining of the mountain) and project the points on to the planes. This way the software can recognise the depth and position of the points.

After this, when fixing the lens the results were the expected.

• 

• 

Both results were made in brute force and everything was exported to jump to maya and finish this shot.

This week was about continuing last week’s assignments from matchmove and lighting with a catchup session from KK and Dom to see how everything was going and make some questions.

Matchmove

After being able to finish the shot on 3DE before my session with Dom, we began by checking if everything was correct and jumped to maya to set up the scene. We imported the mel script and got the scene in the right place, the road points centered on the grid and scaled it up to now build the road.

On wireframe mode, viewing the camera we got from 3DE, I created a plane and adjusted its size to the road from the footage and repeated this process 3 more times. In the end I had 3 separate pieces of road, so I did a bridge on them and adjusted it’s vertexes to fit the road from the footage, this is how it looked like on maya viewport:

Dom showed me a trick for when we have a drone shot, for this we went back to the 3DE scene, tracked a single point on the car and exported “2.5D points to Maya”, this created a mel script that we import to maya. On the maya scene we created a locator and constrained a sphere, as a test, to the locator and this made the sphere follow the car throughout the road.

After this I imported the car given to us for this exercise, to the scene and draged it over the locator, now the car follows the road. I still had to make a few adjustments with keyframes for the car to actually follow the road and this is the final playblast of the tracked shot:

Lighting

For the lighting session I prepared the 2 maya scenes with the chrome spheres. This week KK showed me how to prepare the HDRI (high value and low value) on nuke and how different HDRI can be exported and used in different skydomes to have more control over those values. On Nuke, with both daylight and night light HDRI image, with a clamp node we make every value above 1 set to 1 to discover the low value and with a merge node (on minus option) we can get the high value. This process is the same for both images but ofcourse, need some tweaking in the values and ussually we will need multiple nodes to get to the desired result. In the end we need to export the different values. I had trouble finding out the high value and that is something I will have to ask KK next session.

On Maya, KK showed me how to set up the render layers and told me that, in production, it is best to only use one maya scene with every light in it and use the render layers to get the different lights from that shot. After the session I set up everything that KK showed me and next session we will see continue setting up the lights.

As we got the test shots to track and scenes to light, this week we had to begin working on them, I set a meeting with Dom and KK even before working on the shots as I was sure I would have doubts, specially with lighting.

Matchmove

I began with matchmove and tracked a shot, but only the road and I was getting some errors so Dom told me to track the rest of the scene and after that it was much easier to calculate results. Although, when I had the meeting with Dom he explained a few steps I should follow when calculating the lens distortion, as well as tweak the deviation browser for a better result.

Before the deviation browser this was my result from “calc from scratch“:

This was the “calc from scratch” result after turning off some points from the camera point group:

So I set another meeting with Dom as I was not getting the results I expected. But the session is only next week and I wanted to go through the process after 3DE, in MAYA which is the part that I have more doubts with. So, I jumped the deviation browser process as I was getting better results from not touching it and began tweaking the lens and follow the steps Dom gave me and again, the results were not what I expected. I made a lot of tests and none of them was working, these were the results I was getting:

For lens distortion, when we don’t know the lens that recorded the footage we need to open the parameter adjustment window and calculate distortion and focal length on adaptive mode, then calculate curvature x and y and quartic distortion on brute force, finally when we are happy with the results we add all the parameters and calculate them all to brute force. These were the steps Dom gave me, when I asked for help he told me to calculate only on brute force and these were the results:

When calculating focal length and distortion the expected result is a curved plane, when calculating quartic distortion, focal length and distortion, in the end, the expected result is a cube.

After this we export the maya .mel file, run warp4 to export distorted footage and we are ready to jump to MAYA. With Dom’s help I was able to get everything ready before our session.

Lighting

My session with KK was to understand better the concepts involved in the lighting process, like the environment we want to replicate and what aspects to consider, after KK explained everything he asked me to make a study with bullet points for the car and the cafe exterior both in daylight (overcast) and night light with reference images and gather HDRI for the study.

Street (café)

Characteristics of overcast daylight:

• No direct light source
• Shadows are less strong
• Can’t see much inside the cafe
• Colors are brighter as there is a “general” lighting because of cloud reflection
• Strong Reflex of sky on windows

Characteristics of night light:

• Café lights on
• Lights inside of café are visible
• Light is warmer due to café red color
• Darker shadows
• Darker environment, only light source is un natural and easily changed
• light reflex on floor if wet

Car

Characteristics of overcast daylight:

• Strong reflexes of suroundings on car
• Strong shadows under the car, not around it
• No light on car lights
• Unable to see inside due to reflex
• Bright rims
• Bright car color

Characteristics of night light:

• Strong reflexes of city lights
• able to see inside the car (if there is light inside)
• Color of exterior lights noticeable as it reflects an artificial light to some parts of the car
• Headlights turned on as well as plate light
• Street lights mirror on the car

• 

• 

HDRI examples – daylight (overcast)

HDRI examples – night light

I asked KK if we could do a test next session so he asked me to prepare 2 maya scenes with 2 spheres (chrome and lambert materials) too.

The collaborative project this term was, as I said before, self-directed study so we have to either get a project within the university or outside of it or we could ask Luke for a role in an Indie Film Production. As I did not have any ideas or projects I would like to join I asked for a role in the production, at first I signed up for animation and matchmove but Luke asked me if I could do some lighting shots, so I set a meeting with KK to understand lighting a bit better.

I have to admit, I’m terrible at it but I hope to get a bit better, I should know more about this subject even to light my shots instead of struggling with it each time, it is an important thing to consider if we want to get some emotion out of it.

I will not be doing animation, instead I will be doing lighting and matchmove. Before doing the actual shots we were given some exercises as a preparation for our roles, the lighting task will be getting some day light shots into night light shots as well as light up a crash scene (buildings, car, street) also in day light and night light.

Crash Scene:

For the matchmove role, we have a few shots from pexels, with streets and cars on a street from diferent angles – we need to track them.

• 

• 

• 

• 

•