+1 tag for Google+

Sunday, 15 July 2018

Great Article: Matthew Merkovich Says To Stop With The Crazy Tracking Markers

Matthew Merkovich Says To Stop With The Crazy Tracking Markers

Today I came across a great interview with Matthew Merkovich on LesterBanks dated 2 years ago.

He's a very experienced CG artist. Having spent many years doing matchmove, he gave some of his insights into the quirkiness surrounding the craft and many misunderstood practices that may not be applicable given the capabilities and accuracy of the software tools available to us in the present day.

One of the points he raised is relating to the practice of placing red/orange crosses for markers during location shooting simply because they are told it is the best way to do things.

On top of this, in the interview he also has a list of matchmoving dos and don'ts, advice and tips!

Matthew also has a series of videos showing many matchmove tutorials. Follow him on Vimeo!

Tracking Marker Guidelines from MattMerk on Vimeo.

As an artist that has been doing matchmoving for no small number of years, I am gladdened to have come to agree with many of Matthew's advice.

One of the things that I hear many a co-worker ask for is on-set data, camera info sheet, etc. For many years large and structured VFX facilities have relied on them because they are meticulously taken and recorded down, especially on very large productions. We only get this on the most organised shoots and even then human errors are commonplace when the person measuring it does not fully understand how it needs to be used downstream

One of Matthew's observations I find interesting is that matchmove software of today are already capable of getting accurate solve without needing much camera-specific information. Sometimes errorneous camera readings and information get in the way of the solve instead of helping it.

This scenario is a commonplace-occurance in today's VFX production. Not all location shoots will come back with reliable per-shot camera data and measurements that big pipelines require for their plates to push through.

Thus as an independent freelancer and as an artist working on a smaller scale production nowadays, almost every matchmove project I encounter does not have adequate camera data, if at all. I find myself not needing to ask for camera data. You may think this is an extreme case but it is not. This is actually commonplace.

However, saying that the matchmove process does not require any camera information is probably slightly misleading. I think it's always safe to have camera data, however accurate. Afterall, matchmoving is replicating camera attributes and movement. I see camera data as a back-up and as an instrument to confirm and affirm, for very tough shots when too many variables are involved, when any known variable from data taken from the set would eliminate guesswork and help to greatly speed up the matchmove process.

Foundary Post: How Data Analysis is Improving Matchmoving

Today I came across a post on Foundary's website about data analysis and matchmoving.
Alaistair Barber
Alastair Barber tries to improve and speed up the pipeline with data analysis and algorithms while working with DNEG.

Matchmoving is something close to my heart. I started my Hollywood film career as a matchmove artist. I'm also quite interested in machine learning, data analysis and artificial intelligence. Thus this article grabbed my attention.

It seems that DNEG is a great choice for data analysis, having accumulated 20 years of production data.

The article does not have specific mention of how the result of the analysis is helping to speed up the process of matchmoving, but we know for sure that something is happening in the field of data and the VFX production process.

Data sets and algorithms in the pipeline
Graph source: https://www.foundry.com/trends/business/matchmoving-big-data

However there is one concrete thing I find fascinating in the article. This graph in the article actually gives an overview of the time and resources taken up by each stage in the VFX process.

Monday, 19 March 2018

My Works: TV Commercial

Here's a TV commercial I was involved in last year, while at VHQ. This TV commercial was for JD.com. Almost the whole of the TV Commercial team and Film team were involved.



I helped in look-development of Cyborg, especially in texturing, shader creation and lighting the asset. I was also involved in tweaking textures and shaders for Aquaman's asset.

Also, I helped in the texturing, shader and look-development of the swordfish and the car that got crashed.

For texturing, I used Mari. Rendering was done with Arnold in Maya.

Sunday, 22 October 2017

Maya nParticles Intro - part 02 - nParticles Filling a Cup and Pouring Out



This is a continuation of my Maya nParticles introduction. In this video, I talk about the workflow for: 1. how to create nParticles filling up objects, 2. how to get nParticles to react and collide with animated geometry 3. how to convert nParticles to mesh geometry.
By the end of the lesson we will have a cup filled with particles, reacting to a stirring object, and then having the cup tilt over and pour out the particles.

Saturday, 21 October 2017

Maya nParticles Intro - part 01 - nDynamics Objects Interaction



In my latest video I give a very brief introduction to Maya's nDynamics system.

in it, I show how straightforward it is to set up a scene that contains an nParticle system, an nCloth object and an nRigid geometry, and have them all interacting with each other.