Where i work our bread and butter is in the production of computer generated imagery (CGI). Fully CG or partially CG images of commercial or residential buildings yet to be completed.
These images are made up of 4 key aspects; composition, the scheme, lighting and assets. All with the final goal of advertising to potential buyers.
Its an interesting environment to develop for, code and computer generated imagery. This series is focused on the management of our assets, current pitfalls and frustrations , and finally how we went about developing solutions.
When I say assets i’m mainly referring to 3D models with textures that can be placed into a 3d modelling environments and rendered to a high resolution, realistic image. A 3d asset could be a table, car, street lamp. Honestly, anything.
|A render of a 3d asset courtesy of Model Plus Model.||A render of a 3d scene full of assets courtesy of The Boundary.|
The library is filled to the brim with various assets. This series of posts will only be discussing the type of asset described above, the 3d asset. But for background, other types of asset include (but not limited to) 2d imagery, footage and light information (.ies).
The library is accessed via local-only networked storage. Each office has a different NAS in place hosting their own libraries.
We categorize 3d assets using a few different criteria.
|Typical folder structure of a model categorized by its type.|
The interface to the folder structure will be whatever the 3d modelling package used, supplies. 3ds Max from Autodesk for example, uses a standard Windows OS explorer dialog.
99% of the time these assets will be imported directly into the 3d modelling package. This will usually import the geometry completely with symbolic links to any related textures.
The current library - even with its quirks, works. It's OK. It doesn't take too long to grok and its easy to update. But its old and bloated. More than ready for the eternal slumber. Unfortunately we needed more, more to justify spending the time on it. It was still breathing, so it was still working.
After a bit of napkin math and some more than biased testing conditions involving a stopwatch and a list of items to find in the library.
We came to the conclusion that it takes approx 6 minutes to find an asset. Or, not find an asset, rather confirm we have it or something like it. With n amount of assets per image and a desired n amount of images to complete per month. The number was quite a bit higher than we had estimated.
Unfortunately i couldn’t find any numbers online to compare such a niche query so I had to instead take the results as a sort of baseline.
Any solutions found - at minimum - should shatter those numbers or it would be a waste of time.
With the underlying goal of drastically reducing that 180 hour number - which would be a win on its own - We collected regular users (of the library) frustrations.
This lead to problems that can only been seen from a 1000 foot overview of the entire system.
Something that's ¼ of our deliverable shouldn't be left to rot.
All of the above took a couple weeks. It was mainly driven via coffee fueled lunch breaks and between projects, but it left us with a greater understanding of our current system and a solid list of user stories that we could use as criteria to help with the upcoming decision making process.
Upgrading the library will take time. We understand this and are in no immediate rush. Our current system has been in place for +7 years. If a new system has anywhere near that longevity we’ll be more than happy to send a little more time in the planning stage to ensure we produce a stable system.
This has been part 1 of an ongoing series about upgrading our asset library with the ultimate goal of reducing time-to-deliver. Thanks for reading. If you’d like to notified of the next post in the series, feel free to join the mail list!
If you have any questions regarding the site or any of its content, I'd love to hear from you!
Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur.
For more information or to ask anything at all, email@example.com.