| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
pressing the m key.
|
|
now thrown randomly across each triangle instead of only at the center of each triangle. all of the rendering kernels have been rewritten so that they operate additively; for example, you may now throw photons from the light source onto the scene, render from the camera to the scene, then throw more photons and render again.
|
|
|
|
acrylic vessel. There are serious issues with this model which
probably make it unsuitable for photon propagation at the moment.
(Example: Why do you need to traverse 8 surfaces to cross the acrylic
vessel? There should only be 4.)
|
|
|
|
to integers in alpha blending until the last step.
|
|
|
|
|
|
|
|
|
|
PMTs in the right place and acrylic vessel. No new material properties
yet.
|
|
the hybrid monte carlo ray tracer in a separate thread. the camera object is initialized by passing a pycuda.driver.Context object and a threading.Lock object; you can then run kernels and copy arrays to and from the device in the same context as that used by the camera by acquiring the lock, calling context.push(), executing the kernel and/or copying arrays, calling context.pop(), and then releasing the lock. fixed mistakes in both build_rgb_lookup() and render() where I had accidently switched the green and blue photons. updated the lbne geometry with the latest specifications. added profiles for the sno pmt and concentrator.
|
|
tracer and photon simulation can be run in the same context. added the ability to add alpha channel to triangle color so that triangles can be made transparent. added __noinline__ modifier to certain device functions to speed up kernel compilation.
|
|
you to copy photon information within a thread but still keep a single random number generator throughout the thread.
|
|
of each triangle. reduced the number of runs to average when propagating photons from each pixel in render.py from 5 to 1; the speed improvement outweighs any small improvement in the quality of the rendered image.
|
|
a boundary
|
|
src/photon.h so that photon propagation by propagate() in kernel.cu and the hybrid monte carlo ray tracing use the same code. instead of a single state, photons now carry the history of the processes they've undergone. this history is stored as a bitmask; see src/photon.h. start_node and first_node of the mesh are now stored as global variables in mesh.h instead of being passed to kernel functions.
|
|
on a profile of the PMT model (see build_pmt() in solids/pmts.py). triangle intersection now allows one of the two coefficients multiplying the vectors which span the triangle to float slightly negative (up to -EPSILON; EPSILON is defined in src/intersect.h) in order to eliminate rays passing through the line between two triangles. cleaned up a lot of unused code. pulled duplicate code in view() and render() into functions in view.py. in order to allow view.py and render.py to search pre-defined geometries, solids, meshes, etc. without requiring them to be pre-built, pre-defined geometries, solids, meshes, etc. should be returned by a function tagged by the decorator @buildable(identifier) defined in view.py, where identifier is a string used to identify the object as an argument to either view.py or render.py. optical materials and surfaces are now defined in optics.py. added an image directory to save cool screenshots.
|
|
|
|
used. GPUThread objects now shallow copy the geometry so that threads are not writing to the same memory when the geometry is loaded onto the gpu. the model number for the 12" Hamamatsu PMT is R11708, not r7081 (which is the model for the 10" PMT); all references to the 12" PMT were changed accordingly. only allocate space for 20 materials and 20 surfaces on the gpu instead of 100 to save some space. started to modify track.py to build its own photons and module since the GPUThread object only copies photon hit times back from the gpu (not track information), but I am waiting to find out if pycuda GPUArrays can be used with vector types.
|
|
with only one child.
|
|
improvement.
|
|
it is too slow. Now we do it directly.
|
|
Water Cherenkov case study.
|
|
|
|
|
|
|
|
unique materials and surfaces in each solid.
|
|
|
|
|
|
solid shapes. tried to speed up geometry.build() by replacing some for loops with np.fromiter() and imap().
|
|
threadtest.py to make it easier to flip between different models.
|
|
This avoids the texture size limitation.
|
|
Modify LBNE class to take a cut_pmt option to switch to cut PMT model.
Add corresponding lbne_cut, minilbne_cut, microlbne_cut detectors.
|
|
behind the PMT equator to save memory.
Note: To repair a PMT produced by Sketchup, run Netfabb, start repair
mode, and run the "Remove degenerate faces" action, and apply repair.
Now you can export the part and OpenSCAD will be happy.
|
|
channel, which is presumed to trigger that channel. Major speed up in conversion of detection times to time PDFs.
|
|
|
|
on the GPU. you can now take a screenshot of an image rendered with view.py() by pressing the f12 key.
|
|
sketchup.
|
|
returned from generate_event()
|
|
initializes ALL of the states.
|
|
|
|
geometry.py and moved instances of these classes into separate folders. the Solid object no longer contains a rotation, displacement, or id variable; instead, they are passed to a geometry object when calling add_solid().
|
|
the 12" hamamatsu and sno pmts. ratdb.py is able to parse ratdb files. chromaticity.py provides a function to map wavelength -> rgb color. lbne detector model now includes an outer black cylinder and pmts with a glass layer and photocathode/reflective surfaces.
|
|
scattering, reflection, and refraction
|
|
datatypes everywhere, and build final mesh without concatenation of
lists. This allows for very large detectors, like full size LBNE.
|
|
|
|
textures. also, forgot to include the gpu code for material/surface structures
|
