TESTSHADE_DSO(1)

📦
version 1.15.4.0 · 4 versions
, in fedora 44

🌸 NAME 🌸

testshade_dso — Test Open Shading Language ✨

ok so quick vibe check: this is page TESTSHADE_DSO(1), it lives in the User Commands section of the manual, and it's about a program you actually call as testshade. bestie it is SO good. 🫶

🎀 SYNOPSIS 🎀

testshade [options] shader...

like, you hand it some shaders, you hand it options, it does its thing. iconic behavior honestly. 💅

💖 DESCRIPTION 💖

testshade — Test Open Shading Language. Copyright Contributors to the Open Shading Language project. that's the official line and I'm keeping it, because credit where credit is due, no cap. 🫡

it's the test harness for Open Shading Language — you poke it, it shades stuff, and it tells you whether everything is slaying or not. and the flags?? so many flags. we love a configurable queen. let's go one by one, bestie, I got you:

  • --help — print the help message. first thing you type, always. 🙈
  • -v — verbose messages, so it actually tells you what's going on instead of being mysterious.
  • -t NTHREADS — set the thread count. default = 0, which means auto-detect #cores — it counts your cores for you, main character energy fr fr.
  • --optix — use OptiX if it's available. if it's not available, well. it's not available.
  • --debug — lots of debugging info.
  • --debug2 — even MORE debugging info. yes, more than "lots". we escalated. 🔥
  • --llvm_debug — turn on LLVM debugging info.
  • --runstats — print run statistics. numbers!! ✨
  • --batched — submit batches to ShadingSystem.
  • --vary_pdxdy — populate Dx(P) & Dy(P) with varying values (vs. uniform).
  • --vary_udxdy — populate Dx(u) & Dy(u) with varying values (vs. uniform).
  • --vary_vdxdy — populate Dx(v) & Dy(v) with varying values (vs. uniform).
  • --profile — print profile information.
  • --saveptx — save the generated PTX (OptiX mode only).
  • --warmup — perform a warmup launch. stretch first, bestie. 🤸
  • --res XRES YRES — set resolution.
  • --options LIST — set extra OSL options.
  • --texoptions LIST — set extra TextureSystem options.
  • --colorspace NAME — set ShadingSysem colorspace.
  • -o VARIABLE FILENAME — specify an output (filename='null' means don't save).
  • -d NAME — set the output data format to one of: uint8, half, float.
  • --print — print values of all -o outputs to console instead of saving images. console girlies rise. 💅
  • --groupname %s — set shader group name.
  • --layer NAME — set next layer name.
  • --param NAME VALUE — add a parameter (options: type=%s, interpolated=%d).
  • --shader SHADER LAYERNAME — declare a shader node.
  • --connect FROMLAYER FROMOUTPUT TOLAYER TOINPUT — connect two shader layers. matchmaking but for shaders. 🥰
  • --reparam LAYERNAME PARAMNAME VALUE — change a parameter (options: type=%s).
  • --group CMD — specify a full group command.
  • --archivegroup FILENAME — archive the group to a given filename.
  • --raytype %s — set the raytype.
  • --raytype_opt — specify ray type mask for optimization.
  • --iters ITERS — number of iterations.
  • -O0 — do no runtime shader optimization.
  • -O1 — do a little runtime shader optimization.
  • -O2 — do lots of runtime shader optimization.
  • --llvm_opt LEVEL — LLVM JIT optimization level.
  • --optix_no_inline — disable function inlining when compiling for OptiX.
  • --optix_no_inline_layer_funcs — disable inlining the group layer functions when compiling for OptiX.
  • --optix_no_merge_layer_funcs — disable merging group layer functions with only one caller when compiling for OptiX.
  • --optix_no_inline_rend_lib — disable inlining the rend_lib functions when compiling for OptiX.
  • --optix_no_rend_lib_bitcode — don't pass LLVM bitcode for the rend_lib functions to the ShadingSystem.
  • --optix_no_inline_thresh THRESH — don't inline functions larger than the threshold when compiling for OptiX.
  • --optix_force_inline_thresh THRESH — force inline functions smaller than the threshold when compiling for OptiX.
  • --optix_register_inline_funcs — register functions that should or should not be inlined during LLVM optimization.
  • --entry LAYERNAME — add layer to the list of entry points.
  • --entryoutput NAME — add output symbol to the list of entry points.
  • --center — shade at output pixel 'centers' rather than corners.
  • --debugnan — turn on 'debug_nan' mode.
  • --debuguninit — turn on 'debug_uninit' mode.
  • --groupoutputs — specify group outputs, not global outputs.
  • --oslquery — test OSLQuery at runtime.
  • --print-groupdata — print groupdata size to stdout.
  • --inbuffer — compile osl source from and to jbuffer.
  • --no-output-placement — turn off use of output placement, rely only on get_symbol.
  • --shadeimage — use shade_image utility.
  • --noshadeimage — don't use shade_image utility.
  • --expr EXPR — specify an OSL expression to evaluate.
  • --offsetuv UOFFSET VOFFSET — offset s & t texture coordinates (default: 0 0).
  • --scaleuv USCALE VSCALE — scale s & t texture lookups (default: 1, 1).
  • --userdata NAME VALUE — add userdata (options: type=%s).
  • --userdata_isconnected — consider interpolated=1 to be isconnected().
  • --locale NAME — set a different locale.
  • --use_rs_bitcode — use free function bitcode Renderer services.
  • --jbufferMB JBUFFER — journal jbuffer size in MB.
that's the whole list, bestie. all of it. nothing skipped, because I would never do that to you. 🫶

and that's the page!! testshade_dso 1.15.4.0, dated May 2026 — love that for us. go forth and shade, you gorgeous genius. 🦋✨💖

📦 where this page comes from