Two weeks since reconstruction mini-workshop, so not many things:
TPC plate rendered via textures, ~2 times faster
some fixes in kinematics rendering (but decay point still problematic)
GUI improvements:
lens representation of nest-layouts
fully cutomizable class-views
Will not repeat the whole story (http://agenda.cern.ch/fullAgenda.php?ida=a051718), but:
intermediate representations: digest files and in-memory structures
combine data for selection / analysis
VSD — Visualization Summary Data
pure ROOT .vs. AliROOT mode
Fiddled with the event-loading code yesterday ... SEGVs possible.
load and play with it (navigate, modify render style)
import by regular expression
import with grouping by common volume
interface to overlap checking
export simplified geometry
save visual representation of geometry
volume selection, colors
recursive expansion
encapsulation of whole geometry sub-trees
photo-realistic (or more beautiful) rendering
Nest navigation: reading the structure, list expansion
opening and closing of class-views and nests
Correspondence with implementation: class structure
Nest layouts
Pupil navigation: mouse & key bindings
moving and looking around: M1 & M2 (use the Ctrl button!)
selection of viewing transformation
Rendering Modifiers or RnrMods
render driver provides stack interface for setting properties
e.g. changing line width of a geometry sub-tree by ZGlBlending RnrMod
Gled docs: http://www.gled.org/docs/
Gled for the impatient: http://www.gled.org/docs/html/gledimp.html
Particle selection: TParticle + decay point
Particle Nest layout, Picking, Importing daughters
Removing unneeded objects
Changing the data source
Further options for particle selection
Modifying rendering parameters: ZParticleRnrStyle
Import hits via ZParticle
Hit selection:
Float_t x, y, z; UChar_t det_ID; Int_t particle_ID, eva_ID;
RnrMod example: point size modification with ZGlBlending
Optimized for memory usage: load from tree, store into GL textures and push to graphics card
Modifying rendering parameters: TPCSegRnrMod
somewhat slow
setting the time slice (also need slice thickness)
missing: show time graph for pad / pad-row
Comparing with hits: setting the Near/Far clipping plane
Implement application state with locks against concurrent operations
Close-up presented functionality, build against current AliROOT ... deploy binary release
Digits of other detectors
Clusters / space-points
Reconstructed tracks
Combined representation for cross-searches
At this stage anybody willing to try-it-out is a contributor (and will get lots of appreciation, support and tea/coffee/beer)
Might be some problems with graphics-card drivers (NVidia & ATI drivers must be downloaded from their site)
Kinematics: decay-points, end-of-tracking
Together with some common hit/track-ref representation: could be used as a general Virtual Monte-Carlo visualization suite
Intermediate data structures and converters
Where to put the code ... how to access it from the EVF
Thread-safe TFile operations on ROOT level
Dropped the problem during last 3 months ... but urgent.
Gled source integration: provide scripts for downloading and building of core Gled (a lot of it half cooked for Gled distribution building)