From: Frieda S. R. <fr...@ns...> - 2005-04-08 16:04:56
|
This continues a previous discussion on how to center a subset of atoms without having the structure jump suddenly to a new position. Frieda wrote: > When you center a subsection of a structure as follows, > > center (protein and 1-12) [assuming the protein is > 12 > residues] > > two things happen. One, the structure will now rotate about the > centered set (desired behavior) and two, the structure jumps so that > the centered atoms are at the center of the screen (undesired) with > zoom = 100 (default zoom, undesired). > > This causes the structure to jump abruptly to a new position on the > screen. > > In Chime, "Set zoom center false" disables the second part of these > two actions-- the molecule will rotate about the new center, but does > not jump, it stays put. For continuity, this is much better. Miguel addressed the issue: > I have a prototype operational, but the results are unacceptable. The > problem is caused by 'perspective depth' calculations. That is, if we > 'set > perspectiveDepth off' then everything works OK. However, with > perspectiveDepth turned on (the default) then the perspective view of > the > model changes when the center of rotation is changed. That is, because > the > new center of rotation is not the same as the old one, the view is > different. You can think of the difference as moving the focus of the > camera ... it is almost the same, but not exactly ... so there is > movement. Frieda replied: >> I would like to see an example of this, if possible. I can envision >> what you mean, but how much movement is there? -- perhaps it is an >> acceptable amount? Sometimes small defects such as this can be hidden >> (i.e., human eye distracted) by the move or moveTo command, which is, >> after all, the next thing that is going to happen. >> Miguel: >>> Frieda, >>> >>> I added a switch to enable this change in 'center' behavior >>> >>> set frieda on Note: The Jmol command "set frieda on" has the same effect as Chime's "set zoom center off." My findings: With perspective depth on, and set frieda on, there is still a jump, but it is much smaller than with "set frieda off", i.e., the default state of Jmol. As Miguel explained above, with perspective depth off, and set frieda on, there is no jump whatsoever. But this is not a good solution because perspective depth is a desired characteristic most of the time. If the command immediately following the centering of a subset is a move or moveTo command, the small jump (with perspective depth on) is disguised very well- nearly undetectable, especially if one is not specifically looking for it. One caveat- if you slow down the move command by increasing the time factor (# seconds over which the move takes place) the jump becomes noticeable and distracting. I have tested this with a small (12 residue) subset of a medium-sized pdb file (300 residues), and both small (12 residues) and large (300 residues) subsets of a large file (1400 residues) with favorable results as long as the following move command timing is set to 1 second (default setting)- very hard to see any jump. So for my immediate purposes this fix appears to solve the problem of a jumping molecule when centering a subset of atoms, because I routinely follow with a move command. However I suspect this fix may not be robust enough for all users. Perhaps others who need this behavior could test it out to see if it answers well for them in different contexts? For example, does anyone out there center a subset and then *not* immediately move the centered set to the center of the applet, using a move command? While testing I found a zoom bug that is unrelated to the zoom center issue (since it happens with "set frieda off" in effect): At a high level of zoom, atoms are displayed as 50% spacefill suddenly start to "fuse" so that they look as if they were all in the same plane. A backbone display also behaves similarly, with new, fused bonds appearing. This may have to do with the problems Miguel has mentioned regarding raising the zoom level, I suppose, triggered by using a non-rectangular applet (where zoom is anchored to the larger dimension). Frieda ************************************* Frieda S. Reichsman, PhD Molecules in Motion Interactive Molecular Structures http://www.MoleculesInMotion.com Shutesbury, MA ************************************* |