Edit - added lots more data to the end of this old thread. Also edited 'Optimum' to 'Optimax' - because there's a useful 'Optimin' too.
I have a discovery to offer, after running numerous Alignment cases with a small number of test-sets of photos and picking out ratios etc in Alignment Settings, Alignment Report and Cache file.
Others may know this, whether precisely or intuitively – here it is anyway.
It’s a way to very quickly discover the optimum Alignment Settings for any given photo set. That is, Optimax and Optimin benchmarks which can then be tweaked not at random, by ‘experience’ or by trial and error, but in a calculated way.
Open RC in one window and the Cache folder in another.
Delete everything in the Cache folder.
In 1Ds, Add the photos.
In Alignment Settings, set Max features per Mpx and Max features per image ridiculously high, like 500k and 1000k respectively, so that neither will restrict Points count or Total projections.
Run Align Images.
Look at the Cache folder – if you had say 29 photos you will have 29 Cache files plus one small data file.
Note the MB of the largest file, denoting the best, most detailed and useful photo in the photo set – say it’s 8.2MB (unfortunately the randomly generated filename doesn't tell which photo it is).
Divide it by 62B – 8.2MB/62B = 132KB approx.
132KB is the Max features per image Setting that is just sufficient to capture, without restriction, all of the Features which RC is capable of extracting from the best photo in the photo set (Optimax).
Divide it by the mpx of the camera – say 12.1Mpx - 132/12.1 = 10.9KB approx.
10.9KB is the Max features per Mpx Setting that is likewise just sufficient to capture, without restriction, all of the Features which RC is capable of extracting from the best photo in the photo set (Optimax).
In 1Ds, select the best (hopefully only) Component that’s been created.
In 1Ds Alignment Report, note various results figures esp Points count and Total Projections. Both of these are loosely proportionate to the total MB of the Cache folder (the sum of the individual Cache files created).
Delete everything in the Cache folder.
In 1Ds, delete the Component(s) that’s been created. The photos should still be there, not needing to be Added again.
In Alignment Settings, set Max features per Mpx to 10.9KB (Optimax) or whatever you calc’d, and Max features per image to 132KB (Optimax) ditto.
Run Align Images.
Look at the Cache folder – you’ll see exactly the same distribution of file sizes as on the first run, incl 8.2MB or whatever for the largest file.
In 1Ds Alignment Report, the various results should be close to the results you noted before, because you set Max features per Mpx and Max features per image to be just barely high enough for RC to capture, without restriction, all of the Features which it’s capable of extracting from the best photo in the photo set (Optimax).
Again delete everything in the Cache folder and in 1Ds delete the Component(s) that’s been created.
Try different settings for Max features per Mpx and Max features per image.
If either or both are even a little lower, the MB of the largest Cache file and the total MB of the Cache folder will be a little smaller and in Alignment Report the Points count and Total Projections figures will be a little lower, showing that the latter are now being restricted.
But if either or both are a little or a lot higher than Optimax, the MB of the largest Cache file and the total MB of the Cache folder will be unchanged and in Alignment Report the Points count and Total Projections figures will also be approx. unchanged, showing that there’s no extra to be had.
It’s interesting, in Cache, that any lowering of the Cache file sizes takes the form of several (or all) at a top figure (rather than just one at the topmost figure), showing that several (or all) files are now being top-clipped.
When there’s just one solitary largest file in Cache, that means that top-clipping is not happening, and Points count and Total projections are at the highest that RC can get (Optimax), with this photo set.
Where does that crucial 62KB come from? 61-62KB is just the figure that RC seems to consistently create, across different photo sets – 62KB for each feature that RC captures into Cache.
The beauty of this approach is its speed, using only Cache file sizes. It’s not even necessary to allow Alignment to complete, and not necessary to look at Alignment Report in 1Ds. With 29 photos, it takes 13secs for the Cache folder to fill, then you can look at the largest file size, divide it by 62 and you’re away!
Of the other Alignment Settings, I’ve found that only Detector sensitivity affects all the above, during Alignment – but not a lot unless it's a really poor-textured set of photos.
Image overlap has no effect during Alignment when a photo set has high overlap; when it has poor overlap, then this setting does have a small effect.
Preselector features has no effect on Alignment, that I can find.
In all the Alignment cases that I’ve run, Median error and Mean error barely change for a given photo set, whatever the setting, at about 0.35 and 0.45 with good-textured photos, well below the 0.50 target. So if that’s the measure of Alignment quality, then I wonder what most of the other setting are for!
Comments gratefully received, from anyone who checks this out. There may be combinations of Settings which have more result than I’ve found. It’s a great way to gradually discover what these obscurely-named undocumented settings actually mean and do!
And what we should then do, proceeding from this benchmark.
For example, reducing Max features per Mpx below this Optimax, to ‘starve’ the result, ensuring that absolutely all pixels/points/features get hungrily captured before Max features per image is completely satisfied. Personally, I’ve found that has no effect – there are still holes where there should be adequate texture, and the highly-textured areas still grab all the attention!
Please sign in to leave a comment.