Here we go again

That’s a delightfully ambigious post title for a critical topic – the schedule for completion.  Am I reluctant to jinx myself with another prediction of imminent closure? Indeed I am! No more optimistic announcements, just the facts, ma’am.

Last night I got enough data into the latest version of the Scope of Work post production document to get a result, and based on our present team’s power there’s another 405 days ahead. Since I’m only prepared to budget another 45 days to the final editorial and VFX, that’s a bit of a problem – that zero has gotta go! I’m not worried, the kinks in the SW’s calculations should be worked out by tomorrow.

Though the SW calculations are crude, they are based on data – as opposed to wild guessing. As we progress through the tasks, I’d like to update and improve the calculations based on actual results,  eg the SW predicted roto task X executed by team member A would take 4 hours… but it only took 3, etc.

There are two elements – the tasks and the team. The last time I tried this (3 years ago), I thought only about the tasks ’cause the team was just me. The only relevant information was the hours I assigned to each task, a guess fest. A quick analysis of that document is worthwhile.

As the edit was unfinished, including time code IN and OUT points was a huge waste of time. The scene isn’t helpful either, there are only four acts 0-3. The department allowed this list to be extracted from the much larger master inventory (worth finding). The specialty was actually an attempt to assign priority. The description combined both what the scene was and what the changes were, which is confusing and tough to search. The numbers were an estimate of hours based on my very limited VFX experience at that time. Based on a 40 hour week, 297 hours is 2 months of work. Staring at screens for 40 hours a week is a recipe for blindness, insanity and so forth, 20 something hours is more realistic.

Here’s the current structure for the task portion of the SW (open document format and pdf).

The SW document is based on the current iteration of the edit, so there’s no timecode. The edit will continue to evolve. The tasks are given a number. Each shot might have multiple tasks and these tasks would have related numbers, eg 10.1, 10.2 etc. The task number will be a shorthand for assigning tasks to contractors, (the SW Shot should be Number).

The Description refers to the shot that the task operates on and typically includes a camera angle and a dialog excerpt. This is a quick reference for contractors to locate the task. Shots with ADR dialog might not be easy to find once exporting to AE, as audio won’t get included with the clips.

Proposed Changes is the actual task. If a proposed change effects continuity, then a standardized label is used. The red hallway is a change that extends over 20-30 shots, and using a standardized label allows those tasks to be identified within the SW calculations and grouped as a meta task. An intelligent decision can then be made by the director about which meta tasks to queue. I’ll be grouping tasks into meta tasks this afternoon.

Priority is a shorthand for essential/non essential to the story. The original paradigm included three values: 3 = essential to the story OR fixing glaring distractions like production mistakes and continuity problems, 2 = including context and texture, 1 = a self consistent and vivid universe. In other words 3 = good enough, 2 = pretty good and 1 = great.

With only two months remaining to wrap this project, we want to at least achieve ‘good enough’. That means identifying ONLY the essential. Production mistakes and continuity problems are a matter of basic competency and will be either deleted or fixed. We can vacillate over what’s essential to the story, but if the dialog and performance require visual or auditory support, then that’s also essential. I could wipe out the need for such support by deletion and so identifying essential tasks before returning to editorial would be key.

Once the non essential tasks are grouped, the decision to queue can be made based on aesthetics AND schedule constraints.

Duration in seconds of the shot is used to calculate the time needed to make the proposed change. The Duration is multiplied by Complexity (1-3) and Base hours per Second. The change doesn’t always take place over the entire clip, so this usually results in an overestimate. Perhaps an adjustment variable could be included like estimated percentage of clip or (.1-.9).

Complexity estimates how difficult the change is, 1 being simple (placing a 4 corner pin on a static shot) and 3 being very complex (roto with hair and motion blur). I also used the variable to indicate what percentage of the clip would be changed, but this muddles the estimate so I think I’ll implement percentage separately, (see Duration above). Complexity can also represent multiple tasks of complexity 1 all wrapped together as a 3, eg new production, design and compositing. This is not as accurate, but it’s faster than listing 3 tasks of complexity 1 rather than just 1 task of complexity 3.

The base hours per second reflects the average amount of time in hours an artist might take to complete 1 second (24 frames) of a composting task with a complexity value of 1. This assumes that all the elements to be composited are readily available and additional design decisions are minimal. As the team gets under way, this variable can be adjusted to reflect real world results.

Estimated completion (hours) for each task is calculated by multiplying Duration x Complexity x base hours per second. As I am not an ace spreadsheet guy, I had to kludge the calculations, using hours for minutes, minutes for seconds and seconds for millseconds… but it works.

The other half of the Scope of Work document is the team calculations.

There are five members of the roto team so far. I’ve also got a few others around the state and internationally who I am considering. I want to be able to add folks until the scope of work fits within the schedule (and stays within the budget of $3k/month).

Expertise (1-3) is mostly a measure of experience with solid workflows. Knowledge about how to do crazy cool shit is good, but knowing how to do it efficiently is great. The goal is to do more cool shit and have time off for cocktails, enjoy romantic opportunities, not go blind staring at screens. Expertise figures into calculating the rate of the contractor.

Productivity (1-3) is about speed and competency. The fastest member of the roto team gets a productivity rating of 3, (1.0). The rest of the team can either be as fast, 3/4 as fast (2 or .5) or 1/2 as fast (1 or .3). This is based on ongoing reports of how quickly members finish tasks. If the most productive artist is more than twice as productive than the least, the scale can be expanded to 1-5 or 1-7. As a mid-range contractor, I can measure my own speed to calibrate the scale and then see how much faster the pros are.

Ideally, being able to price roto work by the task would enable a truly open employment situation. I’m not quite ready to manage that, tho it could be a pretty awesome social media startup – posting virtual work that eligible contractors could bid on taking, (criteria for who is eligible might include no child or prison workforce – unlikely for VFX contractors now but maybe not in a few years).

Productivity figures into the power calculation and what their rate is.

Availability (1-3) translates into hours per day 1 = .5 hour, 2 = 2 hours and 3 = 4 hours. It might seem silly to have such a low availability as 1, but currently interns can only offer about 3 hours a week. I also know that an average day for me is 4-5 hours on screens – creative and totally focused. The rest of the time I need to move my body, bath, eat, play, have relationships, do routine and extraordinary acts, etc.  Availability factors into both power and rates.

Day power (hours) is the effective hours per day a contractor can contribute, in comparison with the most productive contractor(s). This calculation is intended to project progress towards completion, not just hours of work. Day power is multiplied by 5 for Week Power.

Rate adds a base rate of $7/hour + expertise + productivity + availability. The highest rate currently is 7+9 or $16. Aside from raising the base rate, expanding the productivity scale to 1-5 or 1-7 would bring the high into $18-$20/hr.

The Projection (days) = SUM(Estimated completion hours) / total artist day power.

The projection cost ($) = Total artists’ rate $/day (the total of all the artists’ rates) x Projection days. Note that the hours an artist is paid for could be more than  her hours of power. The most productive folks have matching hours.