Category: Group

This week, I received the refraction pass and now things have started to look better. First, I added the screen which was missing and started giving it defocus and fixed the colours to match the plate.

I wanted a deformation of the footage when it goes in the hologram. Gonzalo helped with it and later, I played around with it and made it better.  

I had to match the colour of the CG to the OG plate so that the walls look like they are in the same room. 

Later, I gave a little shadow to the window frame, so it falls on the glass of the window and this will make it look more realistic.

I added a little bit of blur to the glass floor because the glass was reflecting everything perfectly like a mirror and not like glass. 

Next, I added the fish and the spaceship outside the window and played around with the exposure, and then I added soften to give it a dreamy look. 

The next step was to relight for this, I used the position pass and choose a few areas which needed lighting like the clock which I felt was too dark, the library, then the wall near the hologram and even the floor I added extra blue light. The planet near the right-hand side wall also looked a little dark, so I added some more light there as well.

I also had to animate the lamp flickering light, but the problem was that the CG was so heavy (150GB) that I couldn’t playblast it so I had to render out the whole sequence and then animate. I used the Normals pass to give an even-out colour throughout the CG plate. Then I added a light ramp to give it a sun glow, in the room. 

I then animated the depth of the video so it looks more realistic. This is how it looks.

This week: I started off by precomping the clean plate so that the file runs smoothly. This week, I got the digital matte painting (DMP) file. When I put it into Nuke I couldn’t see the DMP from the window. Later, I realized that I had to lower the opacity of the window. For this, I used the crypto matte node to select the window and then make it more transparent.

While rendering the sequence, the mirror on the window was picking up the light reflection and the CG lighter want to me to lower the reflection in comp. So I used the crypto matte and there I masked it to specular and lowered it. The other problem was that when I try to fix the reflection, the screen on the right-hand side could not be seen, so we had to render out the screen separately. 

When I was making the windows transparent, I realised that the object out of the windows didn’t have crypto matte so I had to tell the CG to render that separately and later marge it.

The team planned that we will keep the walls on the left-hand side and overlay a texture on it to make it look more old. So I again used crypto matte and then used multiply to control the transparency.

Cleaning up the red box was quite tricky, and not as easy as I had anticipated it to be. I thought that I would use a 3d card and clean it up, but the problem was that the table would overlap the box. 

To fix this issue, I had to 3D project it first and then clean the red box and then add the table on top of the 3D card.

The vent was transparent and I had to place a card to make sure that the vent had an off-black background. 

We had to add the university logo to the frame. I used Photoshop to make the logo and added some paint and spots on it. I used new cars to project it on the wall. This had the problem of the over-scanning and I changed it on scanline render. The white and black helped the alpha.

This week couldn’t do much because there was a rendering issue. One of the models had a lot of faces and the rendering was taking 6 days. Now the CG leads have removed the model from the frame and when they tried to render it, there were a few textures missing too. 

But the team gave me 60 rendered frames to play with AOVs and see if the thing was working. 

The first problem was that the render files were not premult and the whole sequence was broken into 4 parts and it didn’t look right at the end.

Later I found out that there was no transmission pass and because of that the video didn’t match up.

Cleaning up the floor was the most tedious job in this whole project for now. It was fine to clean up the walls but the floor had few issues. A few of the tracking marks were in between the tiles and we had to use the paint tool to clean them up. It was really hard to fake the lines between the tiles and the tracking wouldn’t work, so I had to go frame by frame. 

This week I got a lighted plate from the CG team. I started matching the plate with the CG asset giving it the same tone.

This week was a slow week – I didn’t have much to do, I started off by cleaning the table markers. This was tricky to do because the inpaint node didn’t work on it. I had to ask Gonzalo for help, and later he sent a script for fixing that problem.

The comper also had to remove the handle near the stars and the red box at the of the wall, I asked Christos (who is the lead comper) to do it, so that the division of work would be equal.

Next, I got the EXR file from the CG leads of the models. I had to add them to the WIP clean plate to see if the tracking works fine. I had a problem with the EXR files as I couldn’t lens distort the plate and when I did there were artefacts. Then I asked Gonzalo, thinking that I did something wrong, and he was like there was no overscan when rendering. I told the CG lead and she told me she will fix it on the next render. For now, till the problem is fixed I have not added any lens distortion.

I also completed the match move this week, I had to add more points overall and needed way more points on the corner, at the end of the video.

Last week, I was working on the cleanup of the plate and finished half of it. This week I showed my work to Gonzalo. He told us that we had to follow the mail he sent, so will have to only clean up the walls. In the following weeks, we will complete the remaining work. All in all, I had to redo the whole thing. Gonzalo also told us that he will show us how to work on something and then we will have to continue by ourselves.

When I was cleaning up before, I used the roto paint node, but Gonzalo showed us the easier way which was to use the Inpaint node. This made the work finish quicker in the areas where there are no other opticals. if there is another object interfering with the clean-up areas it doesn’t work right.

I have finished doing the cleanup using the Inpaint node, but next week I will have to ask Gonzalo how to fix that problem.

I started working on cleaning up my plate. Gonzalo always said that there are different ways to do the same thing on nuke but we need to find the best and easy way. This week’s task was to find that way. I started off with a 3D projection but this didn’t match up with the plate. Next, I tried 2D tracking which worked the best. Later, I thought of using the smart vector but even this failed. For now, I am sticking to 2D tracking, till I meet with Gonzalo, because I have a feeling that there is a better way to do this.

I have removed a few tracks from the wall and the floor. It’s not the best and I am not happy with the work. I am planning to do a rough clean-up and later work on it and make it better.

This week, we got the plate for the project and started to camera track using a 3D equalizer. This plate was fun to work with because this plate had tracking points and helped to make the footage smoother. At the end of the tracking, I had a problem with the export. When I was exporting the 3D camera to Maya to see if it was fine it had a bad shack. I tried to fix it, by changing the FPS and the film back. 

On the same day, we had a class with Dom and I asked him about it and even he had the same problem. Later that day he said that when rendering it out we should start from frame 0 not from frame 1.

After doing the changes I couldn’t render the camera and tracker to Maya or Nuke. I tried to fix it by watching YouTube videos and reading up about it on Reddit but I couldn’t fix the problem. I asked for help from Marianna and we both tried to fix it for 3 hours and failed.

On the other side, I started cleaning up the reference frame so that next week when I clean things will go more smoothly while tracking. I broke it into 3 parts so it’s easy to clean up.