How you do you determine a realistic deadline?

Posted on 10. Mar, 2010 by Andrew Price in Articles

Nobody likes missing a deadline. Not you, or the client. It’s stressful, costly and looks bad for your reputation.

However, as artists it’s our job to tell the client how much time it will take to complete a project and how much they can expect to pay. But how do we predict the unpredictable?

To get more practice in this department I like to create deadlines for personal projects, regardless of whether or not there’s a need for one. I find this prevents laziness and endless test and adjustments.

My most recent project was a New York city earthquake animation. So before I started the project I wrote down a list of everything I would need to model:

A bedroom (chairs, desk, bed, bookshelf)
Detailed building (multiple floors, smashed windows, broken walls)
A camera mapped city
Smoke
Falling debris

I didn’t have the luxury of working on this full time, so I estimated the project would take 3 weeks to complete.

And how long did it really take?

3 months.

Where did I go wrong? The biggest problem was failing to predict certain issues.

Here’s a list of the most time consuming tasks that completely skipped the planning stage:

1. Learning Time

I had never modeled a destroyed anything before, so it goes without saying that I would need to learn a few things. Learning takes time, a lot of time. When you don’t know how to do something your entire production grinds to halt whilst you stumble around the internet trying desperately to find an answer.

2. Software Issues

I was using Blender 2.5 Alpha. That last word there should have been a neon flashing billboard that spelled TROUBLE. However, this completely overlooked this fact. As a result, I had to deal with dozens and dozens of unpredictable bugs. The most noteworthy bug was a ‘segmentation fault’ that caused the software to crash during rendering. I spent over a week trying to find an answer. The worst part is, I never got one. It was in Alpha stage. Nobody knew.

3. Over-estimating my skill

Everyone knows how fast they work. Or at least they should. But when I was estimating on how long certain tasks would take, I severely overestimated my skill. Tasks such as modeling the inside of the bedroom took a mere day in my head, but when I sat down and actually started working on it I realized a week was more realistic.

4. Video production

Model, Texture, Light, Render = Finished! Right? No. There’s also, re-rendering problematic frames, compositing passes, buying sound effects, exporting in various formats for different media, uploading and distributing. They are all crucial tasks, but they never entered the planning stage.

5. Being a perfectionist

After I spent a week detailing rooms of the building that the audience would only see for a split second, I realized that I was being a little OCD. Every artist wants their work to be perfect. After all, that’s what makes great art. However, there comes a point when you need to step away from the computer and and accept that it’s not perfect, but it will pass.

The project was an enormous learning experience for me. It made me wonder: Is determining a realistic deadline a skill that can be taught? Or is it something that only comes with experience?

At completion I jokingly thought that perhaps in the future I should just triple my original estimate. But on a second thought, maybe that’s even such a bad idea?

I put the question to my twitter followers and facebook fans. Here were their suggestions for determining a realistic deadlines:

@r3dp_01: I breakdown the layers of the project, 3d , matchmove etc. then the level of the artist(s)

@francoisgfx: The first thing they teach me when I started at Ubisoft: Always triple what you think

@DarkCellar: Complexity, team size, commitment, experience, skillset.

@OscarMopperkont: Realistic planning.

@Laxy: I work out roughly how long it’ll take then double it. But that’s me dealing with my optimistic calculations

Renato Sousa: Multiply by two…

Brian Knezevich: By figuring out how the project is coming along. Of course you determine it by the time and the work you all ready have done.

Mark Walder: Depends on the desired outcome, Time, detail, texturing lighting, placement and strength of lighting, HDRI background as part of that lighting and Ambient effects, In my experience it is hard to follow real world lighting so to keep rendering and adjusting has been the key for me, and that can make determining a deadline difficult at times.

So how do you create realistic deadlines? What methods have you found that work for you? Share your experiences and lessons learned in the comments below!

Tags: discussion, production

13 Responses to “How you do you determine a realistic deadline?”

Galvedro

10. Mar, 2010

0. You need to have historic data. I don’t mean “feeling”, I mean data.

1. Do a proper decomposition of the tasks at hand. As you say in your post, you have to take everything into account, including learning, dealing with people, buying things, etc.

2. Use PERT (see wikipedia). Make three estimations per task: (O)ptimistic, (E)xpected and (P)esimistic. From those three, take T = (P + 4*E + O) / 6. Be sincere to yourself on the three estimations.

3. With historic data you can adjust the formula above to suit your own bias. If you tend to be optimistic, you can weight P a bit more, etc.

This method is very effective for me.
Stephs Magic

10. Mar, 2010

TY ! Very inspiring post Andrew (and a great video too). My projects rarely transform from mental storyboards into something more tangible :/ So for my next personal deadline I should probably quadruple my estimate
Lars Scheithauer

10. Mar, 2010

First of all: you can’t. There are certain things that are unpredictable like your problems with blender 2.5, and that won’t change in the future. So it’s important to put in a buffer, that depends on the amount of unpredictable stuff. I generally use 20% buffer in my estimates and add percentages for stuff I don’t know (alpha software: +20%, something you never did before: +30%, etc).

Next most important thing is to break down your project into manageable milestones and actions to reach those. The actions should completable in one workday at best. This allows you to see problems with the deadline a lot in advance and figure out what to do with your client.

Then, use a timetracker to check how much time you actually used. After the project is finished, you have to go over that data again to figure out, where and why your estimate was totally off and how to prevent that in the future. While this task is always annoying, especially when a project went wrong and all you want to do is file it in the archive, you won’t learn anything if you don’t do that step.

Last tip (especially true for artistic work as your field of expertise and perfectionists, which I am, too): use the approach of having milestones, that reflect the state of work from rough to detail. This leaves you with a very good basis in the early stages, does give your clients plenty of time to address issues and communicate changes, and restrains the perfectionist in you to the end of the whole project, where you might have or not have time to give him credit.
This approach does work very well for my students, too. When they miscalculated a project and only finished 1/4th of it to the deadline, the mark will be really bad. But if they used the above approach and only made it to the first milestone, that means they already have something on the whole project and not only on 1/4th. It may not be perfect, but they still made it and thus will get a much better mark.
Craigsnedeker

10. Mar, 2010

Wow, I never really think of this. I don’t have deadlines for myself, I go-with-the-flow (I work fast, but my clients are dreadfully slow replying to emails, which is unpredictable) lol
Mike

10. Mar, 2010

Lately I’ve been limiting my working time to 3 hours a day.
It really focuses you – and you always have lots of sleep.
DarkCellar

10. Mar, 2010

I definitely agree that you have to include extra time to deal with uncertainty.

One way is to take the task list and estimate time for each one, and your level of certainty (as a percentage) in the estimate. If the certainty is too low, say below 65%, then break that task down into smaller parts and estimate for each. That way you can find where your uncertainty is.

-Mitch
balaraj

11. Mar, 2010

I want to start to learn 3D
thanking you
Maria

21. Mar, 2010

In the programming shop where I work they practice “The Scotty Principle.” Based on the confession of Mr. Scott that he always inflated the amount of time is would take him to get something repaired and then looked like a savior when he got it done early. We have learned that the best route when telling the client managers a deadline on the spot is to figure what we think it will take multiply, by 3 for the real amount and then multiply by 1.5 for a nice buffer, looking like wizards and googling.
Franco

12. Apr, 2010

What doesn’t seem to have been addressed yet is client changes and their effect on project duration.

To reduce delivery time, doing only the absolute minimum amount of work required in 3d and getting to a compositor as quickly as possible can save a significant amount of time, specially in the client change rounds.

Once the animation is locked, rendering individual passes for material channels, depth maps, object buffers, and each light, can offer a significant amount of flexibility.

You can, for example, darken one building, change its color, or re-render only a few “layers” like the debris in the foreground. Motion channels and depth maps also help you blur individual layers and add depth of field in compositing, which saves significant rendering time.

This increases project setup time, but decreases overall project and rendering time, specially for the change rounds.

When you’re working with clients, the change rounds can often last longer than the original time required for animation. So setting up project set up from the get-go for flexibility like this can be advantageous.

Cinema 4D is strong in the area, with a built in multichannel renderer that can be opened up as a multi layered AE or Motion project. If blender can do the same, I’d love to see a tutorial on that.

Best,

F
Nixon

08. May, 2010

very interesting post, i fail to estimate time needed for my stuff still but i’ll hope to get better by the time…
is there a’ you’ too much in the headline…
shouldn’t it be ‘How do you determine a realistic deadline?’
Since I’m german I’m not sure tho if i might get the sentence wrong but anyway
Eric Crystal Clear Films

13. May, 2010

Well to start once you figure how long it will take then double it and then add 15% more time
So for example it will take a day then make it a day and a half, and like Scotty from Star trek said “if it will take you an hour the tell them two and a half that way when your done in an hour and a half you look like a genius.”
As far as editing you can figure at least 20min of editing time for every one min of finished product.
Mike

13. May, 2010

Franco, you will be pleased to know that both 2.49 and 2.5x can do multi-pass multi-layer renders to the “MultiLayer” format (which is OpenEXR… although not the same as the “OpenEXR” choice(!)). Exactly as you describe for C4D.

I second your suggestion … and third … and fourth.
a

07. Jun, 2010

cool, but you can’t make a movie???