Helping "incremental" optimization

Bug #816850 reported by Frederic Da Vitoria on 2011-07-27
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Hugin
Wishlist
Unassigned

Bug Description

One of the optimization methods is: triggering each optimization option in turn and accepting the results if the new optimization iteration are better than those of the previous iteration. The only problem here is that you need to remember the results of the previous iteration in order to be able to compare! It would be nice if Hugin displayed the results of the previous iteration too when asking of we want to apply the changes.

Yuv (yuv) wrote :

not very useful: the results you want to compare are visual (with undo/redo) and not the numbers. A lower RMS is not automatically associated with a better outcome.

Changed in hugin:
importance: Undecided → Wishlist
status: New → Triaged
Frederic Da Vitoria (davitofrg) wrote :

First of all, this is absolutely not an important issue to me, just a small improvement which would make my user experience easier.

It is precisely the "lower" which is the issue (or maybe I missed something). Each run gives 3 values, average, standard deviation and max. I remember enough from my university lessons to know that the max is the least important, but between average and standard deviation... So I suggest Hugin could say it more clearly, for example average:### (standard:###, max: ###)

Then there is the issue that the variation can be pretty small, sometimes I did note enough digits from the previous iteration so I am not sure if the new iteration is an improvement or not. I thought (but maybe I'm wrong here) that if an iteration is not an improvement, you might skip it and try directly the next step. This is usually how I do it. Anyhow, if an iteration is not an improvement, and of course if Hugin remembered the results of the previous iteration, it could also advise the user no to apply the "worse" iteration and suggest how he should proceed (skip the step or try to improve the control points).

This discussion helps me getting a better understanding of what the standard deviation and the max mean (I know, this is not the place for it): I guess they are telling the user if the control points are correct or if they need improvement. But frankly, I am unable to tell what values of the standard deviation seem correct and which would suggest the CPs should be edited.

To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers