I’m sure you’ll get different opinions on this one. For me though when I’m doing a real model that could get changed by the client, I save out a version prior to adding fillets and ideally save out multiple incremental versions from concept to final. This is not parametrics of course but rather planing for the inevitable. The final model can also be used to make changes though with practice. For instance, ExtractSrf>Untrim>Join can be used to remove fillets.
Layers are really they key here. Especially when you are wanting to work with a clean environment.
I generally work from left to right copying each time I make a major change. Depending on what you are building/changing it’s not necessary to change the layers for each update but that’s really your discretion as to how many layer variations you want to manage.
Construction curves as I like to call them should ideally be on a different layer. I find they get in the way most of the time.
Fillets I believe you can check with simply dimensioning the object. Fillets are also accessible with some checking. Easiest to just write them down I find. Or you can make notes using the text commands.
It’s a good idea to at least copy the object when you are doing chamfers and fillets in case you need to go back to that original model to change things up.
I build in watertight “chunks” I do a lot of car stuff, so for instance, I do the greenhouse as one “chunk” the lower box as another “chunk” then then bumpers as another “chunk” etc, etc…this allows me to adjust the proportions and relationships between the various components using cage edit, bend, scale, etc… I send several updates along the build and have the client mark up screen shots of my progress. This allows the client to have input the entire way, and iteratively hone in on a solution to the given design problem. THEN when everything is dialed in and proportions are nailed, I start adding transitional surfaces (fillets and blends) and details. This has proven to be a really efficient way to work in a highly fluxing client environment. Thousands of models for the toy biz, and I’m still moderately sane, so something has got to be working…
I have been ‘burned by the booleans’ when the client asks for changes, so I developed a curve-based technique that makes it far easier and faster to modify. I made a nice summary in my recent lynda course in one video, and it’s available for a free preview at the link below
I have 4 main layers - Crvs, Layout, Parts, and Boolean. Sub layers under Crvs save the construction crvs for each part of the final model. Layout saves how the parts are positioned in space, with Flow, FlowAlongSrf, cage editing, etc. Parts saves off a final version of each part.
The last part of it is copying all the Parts to the Boolean layer (I have a sub layer for prototyping and another for render). I do enough boolean work to get it looking correct to show the client and send for approval. If they approve, I finish the boolean ops and send it off for rapid prototype.
If they want a change, I’ve hopefully set myself up the best way possible. Although sometimes, it ends up being a do over anyway
