Units absorbance

Project.toml

@@ -1,7 +1,7 @@
 name = "SwissVAMyKnife"
 uuid = "fafdf525-54c7-48ec-ae29-528bd4cccfab"
 authors = ["roflmaostc <[email protected]>"]
-version = "0.1.1"
+version = "0.2.0"
 
 [deps]
 CUDA = "052768ef-5323-5732-b1bb-66c8b64840ba"

examples/1_glass_vial_pattern_optimization.jl

@@ -1,5 +1,5 @@
 ### A Pluto.jl notebook ###
-# v0.19.38
+# v0.19.42
 
 using Markdown
 using InteractiveUtils
@@ -67,6 +67,12 @@ However, we add a trailing 1 z dimension.
 sz = (256, 256, 1)
 
 # ╔═╡ b2dae3ea-5a40-498f-a134-5a9a98e58de0
+# ╠═╡ disabled = true
+#=╠═╡
+target = box(Float32, sz, (150, 110, 1), offset=(90, 160, 1)) .- box(Float32, sz, (80, 50, 1), offset=(90, 160, 1));
+ ╠═╡ =#
+
+# ╔═╡ 827f6f0a-12d9-4a72-8f8b-c3869a037590
 target = box(Float32, sz, (150, 110, 1)) .- box(Float32, sz, (80, 50, 1));
 
 # ╔═╡ 047023bf-96dd-4427-aa13-3fb2340af90e
@@ -86,7 +92,7 @@ Let's use the `LossThreshold` to achieve this:
 "
 
 # ╔═╡ ae3fdae9-7b3c-4391-8fff-530f7758942c
-loss = LossThreshold(thresholds=(0.85, 0.95))
+loss = LossThreshold(thresholds=(0.9, 0.98))
 
 # ╔═╡ b58c9e1f-7ca6-4171-8004-2b84d33cf335
 md"## Specify Geometry
@@ -99,13 +105,13 @@ As absorption we set $μ=nothing$, so we ignore absorption first!
 "
 
 # ╔═╡ 0dd200a7-a0f0-416e-ad69-7343e210a36e
-angles = range(0, 2π, 400)
+angles = range(0, 2π, 401)[begin:end-1]
 
 # ╔═╡ 303a363f-f941-45b9-9194-62a2c411964a
 μ = nothing
 
 # ╔═╡ 6d130809-8b78-485c-bba1-fc958bc9d873
-geometry = ParallelRayOptics(angles, μ)
+geometry = ParallelRayOptics(angles=angles, μ=μ, DMD_diameter=10e-3)
 
 # ╔═╡ cb8b9ccc-4085-4abe-bd71-5e70cc8decfa
 md" As optimizer we use a gradient descent based variant"
@@ -120,6 +126,7 @@ Let's try to run the optimization: `togoc` moves the target onto the CUDA device
 # ╔═╡ 07e97af6-f30b-4100-999c-c61b52cec76e
 @mytime patterns, printed_intensity, optim_res = optimize_patterns(togoc(target), geometry, optimizer, loss)
 
+
 # ╔═╡ c13abe2d-ff27-42e2-9534-cb1936e68236
 optim_res
 
@@ -148,31 +155,35 @@ Printed intensity ----------------------------- Printed intensity after threshol
 # ╔═╡ 5801f61a-79fa-4326-82dd-3901dd4d493a
 [simshow(Array(printed_intensity[:, :, 1])) simshow(ones((sz[1], 5))) simshow(thresh .< Array(printed_intensity[:, :, 1])) simshow(ones((sz[1], 5))) simshow(target[:, :, 1])]
 
+# ╔═╡ 605e1421-23dd-465d-b94d-3a649154f6f8
+simshow(360 .< Array(backproject(patterns, angles; μ)[:, :, 1]))
+
+# ╔═╡ b08e897d-c260-4a42-8065-989abbbcab4e
+simshow(Array(backproject(reverse(patterns, dims=Tuple(())), .-angles; μ)[:, :, 1]))
+
+# ╔═╡ 44ea2ea2-1e4f-44d3-aa02-68e96d6c7b39
+size(patterns)
+
 # ╔═╡ 92f5b987-13c7-463c-b150-71c51ec4aa72
 simshow(Array(patterns)[:, :, 1], cmap=:turbo)
 
 # ╔═╡ 6654e278-3d71-48a3-9382-7b8e486f4d72
 sum(patterns ./ maximum(patterns)) / length(patterns)
 
-# ╔═╡ 3fb2afdb-2ad2-4ce5-9ca3-e1c5c1d45970
-histogram(Array(patterns[:]), yscale=:log10)
-
 # ╔═╡ 11af6da4-5fa2-4e48-a1c8-32bc686b8239
 md"# 4. Include Absorption of the Photo initiator
 Let's add some absorption!
-
-256 is the size of the object in pixels.
-A $μ=2/(256)$ would mean that after propagation through the vial, the ray has only $I(x) = I_0 \cdot \exp(-2) ≈ I_0 \cdot 0.135$ of the initial intensity left.
-
-And the printing is still possible!
-Empirically, a factor of 3 might be possible to print. A higher $\mu$ means actually shorter printing times.
+$\mu$ expects units of inverse meters.
+Empirically, a factor of 3/10mm is possible to print. 
+This is equivalent that only $\exp(-3)$ of the light is left after the propagation through the vial (if the vial has a diameter of 10mm).
+A higher $\mu$ means actually shorter printing times. 
 "
 
 # ╔═╡ 2f51ec68-09c6-4475-81b9-ec252fba5df7
-μ2 = 2 / (256)
+μ2 = 3 / (10e-3)
 
 # ╔═╡ 2e76c088-e61f-4deb-ac69-b73d6e3f361c
-geometry_μ = ParallelRayOptics(angles, μ2)
+geometry_μ = ParallelRayOptics(angles=angles, DMD_diameter=10e-3, μ=μ2)
 
 # ╔═╡ 0e4da85b-99bd-47e0-a8fe-3c69c3a05708
 @mytime patterns_μ, printed_intensity_μ, optim_res_μ = optimize_patterns(togoc(target), geometry_μ, 
@@ -218,10 +229,10 @@ Luckily, you just have to specify the following parameters for the geometry:
 # ╔═╡ f9a82207-6841-44c9-9aec-e8da3de8e0b0
 geometry_vial = VialRayOptics(
  angles=angles,
- μ=2/256,
+ μ=0/256,
  R_outer=(16.60e-3) / 2,
  R_inner=(15.2e-3) / 2,
- DMD_diameter=14.6e-3,
+ DMD_diameter=16.60e-3,
  n_vial=1.47,
  n_resin=1.4849
 )
@@ -266,7 +277,7 @@ sum(patterns_vial) / (maximum(patterns_vial) * length(patterns_vial))
 md"# 6. Reduce Sparsity"
 
 # ╔═╡ 699a0dbd-df85-41b9-b751-cc936d6c03c5
-loss_sparse = LossThresholdSparsity(thresholds=(0.75f0, 0.85f0), λ=3f-6)
+loss_sparse = LossThresholdSparsity(thresholds=(0.9f0, 0.97f0), λ=500f-6)
 
 # ╔═╡ 006aa5f6-8008-455a-a2a6-54eeb0c097bf
 @mytime patterns_vial_s, printed_intensity_vial_s, optim_res_vial_s = optimize_patterns(togoc(target), geometry_vial, 
@@ -276,6 +287,9 @@ loss_sparse = LossThresholdSparsity(thresholds=(0.75f0, 0.85f0), λ=3f-6)
 # ╔═╡ 9f7d62be-378f-4b3c-bf38-239c9daa8705
 plot_intensity_histogram(target, printed_intensity_vial_s, loss_sparse.thresholds)
 
+# ╔═╡ 93f5bffa-f916-4d4b-90f5-751613b48efb
+calculate_IoU(target, Array(printed_intensity_vial_s .> 0.93))
+
 # ╔═╡ b4c7ed0d-e31e-417b-abaf-2193980c9378
 @bind thresh5 PlutoUI.Slider(0:0.01:1, show_value=true, default=0.5)
 
@@ -288,11 +302,15 @@ simshow(Array(patterns_vial_s[:,:,1]), cmap=:turbo)
 # ╔═╡ dc0c21f2-980a-4c97-9a9d-b3d824121bbe
 md"
 The efficiency can be defined as the mean of the pixel values divided by the maximum.
-Because in experiment the real absolute maximum is fixed by the light source intensity. Hence we would like to maximize the real efficiency of the DMD
+Because in experiment the real absolute maximum is fixed by the light source intensity. Hence we would like to maximize the real efficiency of the DMD.
+In this case the efficiency is $\eta$ = $(sum(patterns_vial_s) / (maximum(patterns_vial_s) * length(patterns_vial)))
 "
 
-# ╔═╡ 23c9a442-d892-464d-9174-960eca8de2e7
-sum(patterns_vial_s) / (maximum(patterns_vial_s) * length(patterns_vial))
+# ╔═╡ a0f85250-7c40-4026-b2b2-d267ec58de6f
+histogram(xlabel="pixel intensity", ylabel="Occurence", Array(patterns_vial)[:], yscale=:log10, title="Without sparsity term")
+
+# ╔═╡ 0a805204-d2ad-4672-88ab-4acdff243083
+histogram(xlabel="pixel intensity", ylabel="Occurence", Array(patterns_vial_s)[:], yscale=:log10, title="With sparsity term")
 
 # ╔═╡ a7cafba6-849a-4eb8-9709-a76cb98e9879
 md"# 7. Let's do a bigger 3D object!
@@ -305,17 +323,18 @@ We deactive the cell by default. However of the (...) next to the cell and click
 
 # ╔═╡ 8824fb56-0fbf-4cba-9aea-6449627923f2
 geometry_vial2 = VialRayOptics(
- angles=range(0, 2π, 300),
- μ=2/256,
+ angles=range(0, 2π, 1000)[begin:end-1],
+ μ=nothing,
  R_outer=(16.60e-3) / 2,
  R_inner=(15.2e-3) / 2,
  n_vial=1.47,
- n_resin=1.4849
+ n_resin=1.4849,
+ DMD_diameter=10e-3
 )
 
 # ╔═╡ 0a655d51-e3b6-413b-83de-9781974242a2
 begin
- KK = 128
+ KK = 400
  target_3D = box(Float32, (KK, KK, KK), (80, 80, 100)) .- 
  box(Float32, (KK, KK, KK), (60, 50, 80));
  target_3D = Float32.(Bool.(target_3D) .|| (rr2(size(target_3D)) .< 30^2))
@@ -327,9 +346,12 @@ md"z slide value $(@bind slice PlutoUI.Slider(axes(target_3D, 3), show_value=tru
 # ╔═╡ a05d8dd4-40c3-41da-9dbd-f58ab161b2d4
 simshow(target_3D[:, :, slice])
 
+# ╔═╡ 1bdd07f2-35df-4d5e-b296-7159ae774aa3
+geometry3 = ParallelRayOptics(angles=angles[1:1], μ=1/16.6e-3, DMD_diameter=16.6e-3)
+
 # ╔═╡ d6a59254-bffe-4116-9335-8884eb44556f
 @mytime patterns_3D, printed_intensity_3D, optim_res_3D = optimize_patterns(togoc(target_3D), geometry_vial2, 
- GradientBased(optimizer=Optim.LBFGS(), options=Optim.Options(iterations=20, store_trace=true)) 
+ GradientBased(optimizer=Optim.LBFGS(m=3), options=Optim.Options(iterations=20, store_trace=true)) 
  , loss)
 
 # ╔═╡ 6009b601-6988-4be8-a519-7c59660c73ab
@@ -341,11 +363,14 @@ md"threshold value=$(@bind thresh4 PlutoUI.Slider(0:0.01:1, show_value=true, def
 # ╔═╡ 10cc0e79-5d5b-43cf-afb8-ad21875d6d97
 md"z slider value $(@bind slice2 PlutoUI.Slider(axes(target_3D, 3), show_value=true, default=0.5))"
 
+# ╔═╡ 2a879d1f-c430-48dd-9264-4c1d6b5bbf18
+exp(-0.99)
+
 # ╔═╡ 0835d1b0-a562-4edf-939e-3044f94e53c7
 [simshow(Array(printed_intensity_3D[:, :, slice2]), set_one=false) simshow(ones((size(target_3D, 1), 5))) simshow(thresh4 .< Array(printed_intensity_3D[:, :, slice2])) simshow(ones((size(target_3D, 1), 5))) simshow(target_3D[:, :, slice2])]
 
 # ╔═╡ 83793ffe-38de-4a76-b80f-8dee1d18c5a5
-plot_intensity_histogram(target_3D, printed_intensity_3D, loss.thresholds)
+plot_intensity_histogram(target_3D, printed_intensity_3D, loss.thresholds, xlim=(0.7, 1.1))
 
 # ╔═╡ 262b96e8-a78c-441c-9b46-2d87636286d7
 md"angle $(@bind angle PlutoUI.Slider(axes(patterns_3D, 2), show_value=true, default=0.5))"
@@ -367,6 +392,7 @@ simshow(Array(patterns_3D[:,angle,:]), cmap=:turbo, set_one=true)
 # ╟─465e830f-ca33-427a-ad7b-e18f3bbaa3c4
 # ╠═4475e0c5-4ae6-403f-bb2b-c5bc8a25d18a
 # ╠═b2dae3ea-5a40-498f-a134-5a9a98e58de0
+# ╠═827f6f0a-12d9-4a72-8f8b-c3869a037590
 # ╠═047023bf-96dd-4427-aa13-3fb2340af90e
 # ╟─1d2bdd44-95d2-45ef-8e24-c8ca3df7485a
 # ╠═ae3fdae9-7b3c-4391-8fff-530f7758942c
@@ -385,9 +411,11 @@ simshow(Array(patterns_3D[:,angle,:]), cmap=:turbo, set_one=true)
 # ╟─dc538ea6-54e3-4a0f-a5ef-427802239cf2
 # ╟─71333d50-fa28-4bdf-bad4-d787682e043a
 # ╟─5801f61a-79fa-4326-82dd-3901dd4d493a
+# ╠═605e1421-23dd-465d-b94d-3a649154f6f8
+# ╠═b08e897d-c260-4a42-8065-989abbbcab4e
+# ╠═44ea2ea2-1e4f-44d3-aa02-68e96d6c7b39
 # ╠═92f5b987-13c7-463c-b150-71c51ec4aa72
 # ╠═6654e278-3d71-48a3-9382-7b8e486f4d72
-# ╠═3fb2afdb-2ad2-4ce5-9ca3-e1c5c1d45970
 # ╟─11af6da4-5fa2-4e48-a1c8-32bc686b8239
 # ╠═2f51ec68-09c6-4475-81b9-ec252fba5df7
 # ╠═2e76c088-e61f-4deb-ac69-b73d6e3f361c
@@ -398,7 +426,7 @@ simshow(Array(patterns_3D[:,angle,:]), cmap=:turbo, set_one=true)
 # ╠═17621d91-1f93-4fcf-abf1-5e97463a0bdf
 # ╠═05c6a38e-a87d-46da-acb6-33b393369f4c
 # ╟─f940c2a6-ebc2-4dca-986c-fe25bfa9e4f0
-# ╟─f9a82207-6841-44c9-9aec-e8da3de8e0b0
+# ╠═f9a82207-6841-44c9-9aec-e8da3de8e0b0
 # ╠═71b479a0-29fc-4411-a02d-096427f2a631
 # ╠═675764cb-721a-4b89-92c4-8a1ab7f5867f
 # ╠═9ddd098d-2d78-4de8-a322-40a2463adcda
@@ -412,21 +440,25 @@ simshow(Array(patterns_3D[:,angle,:]), cmap=:turbo, set_one=true)
 # ╠═699a0dbd-df85-41b9-b751-cc936d6c03c5
 # ╠═006aa5f6-8008-455a-a2a6-54eeb0c097bf
 # ╠═9f7d62be-378f-4b3c-bf38-239c9daa8705
+# ╠═93f5bffa-f916-4d4b-90f5-751613b48efb
 # ╟─b4c7ed0d-e31e-417b-abaf-2193980c9378
 # ╟─afa94e1a-d426-4a29-b0fd-73bcdc3e914f
 # ╠═0fc8abe6-2a22-477e-b2a4-6ccba1e16991
 # ╟─dc0c21f2-980a-4c97-9a9d-b3d824121bbe
-# ╟─23c9a442-d892-464d-9174-960eca8de2e7
+# ╟─a0f85250-7c40-4026-b2b2-d267ec58de6f
+# ╟─0a805204-d2ad-4672-88ab-4acdff243083
 # ╟─a7cafba6-849a-4eb8-9709-a76cb98e9879
 # ╠═8824fb56-0fbf-4cba-9aea-6449627923f2
 # ╠═0a655d51-e3b6-413b-83de-9781974242a2
 # ╟─877b1484-969c-45d7-a3e7-4f0301a81a4b
 # ╠═a05d8dd4-40c3-41da-9dbd-f58ab161b2d4
+# ╠═1bdd07f2-35df-4d5e-b296-7159ae774aa3
 # ╠═d6a59254-bffe-4116-9335-8884eb44556f
 # ╠═6009b601-6988-4be8-a519-7c59660c73ab
 # ╟─fa45b520-18a8-4465-b642-a1c08de48e20
 # ╟─10cc0e79-5d5b-43cf-afb8-ad21875d6d97
+# ╠═2a879d1f-c430-48dd-9264-4c1d6b5bbf18
 # ╟─0835d1b0-a562-4edf-939e-3044f94e53c7
-# ╟─83793ffe-38de-4a76-b80f-8dee1d18c5a5
+# ╠═83793ffe-38de-4a76-b80f-8dee1d18c5a5
 # ╟─262b96e8-a78c-441c-9b46-2d87636286d7
 # ╠═c8774a2b-47e0-4be6-b213-bdef6d7b0726

examples/2_benchy_boat_optimization_ray_optics.jl

@@ -1,5 +1,5 @@
 ### A Pluto.jl notebook ###
-# v0.19.40
+# v0.19.42
 
 using Markdown
 using InteractiveUtils
@@ -112,7 +112,7 @@ loss = LossThreshold(thresholds=(0.7, 0.8))
 angles = range(0, 2π, 201)[begin:end-1]
 
 # ╔═╡ 93cebf3f-389b-4fde-8d13-af3e1165ad9c
-geometry = ParallelRayOptics(angles, nothing)
+geometry = ParallelRayOptics(angles=angles, μ=nothing, DMD_diameter=Inf)
 
 # ╔═╡ 2ffbef8c-066d-49b6-8a29-1ef710099029
 optimizer = GradientBased(optimizer=Optim.LBFGS(), options=Optim.Options(iterations=40, store_trace=true))

examples/4_realistic_pattern_optimization.jl

@@ -1,5 +1,5 @@
 ### A Pluto.jl notebook ###
-# v0.19.38
+# v0.19.42
 
 using Markdown
 using InteractiveUtils
@@ -83,22 +83,30 @@ md"# 2. Specify Optimization Parameters"
 loss = LossThreshold(thresholds=(0.65, 0.75))
 
 # ╔═╡ e5bec833-47c5-4f65-8667-f28d4ccb1fbc
-angles = range(0, 2π, 301)[begin:end-1]
+angles = range(0, 2π, 1001)[begin:end-1]
 
 # ╔═╡ e45eeeed-a86e-40e1-b4a2-e4c25cc6a368
 optimizer = GradientBased(optimizer=Optim.LBFGS(), options=Optim.Options(iterations=20, store_trace=true))
 
 # ╔═╡ f618cc32-6851-42de-b2c2-0ddc70b4668d
 geometry_vial = VialRayOptics(
  angles=angles,
- μ=2/550,
+ μ=2 / 14.6e-3,
  R_outer=(16.60e-3) / 2,
  R_inner=(15.2e-3) / 2,
  DMD_diameter=14.6e-3,
  n_vial=1.47,
  n_resin=1.4849
 )
 
+# ╔═╡ 29bbf20e-da89-42af-b7a0-a921c1c5fe3f
+md"
+In this case the DMD is smaller than the glass vial.
+That means, our DMD area is smaller than the simulated target volume.
+The resulting pattern size is: $(floor(Int, (14.6e-3 / 16.6e-3) * size(target, 1))) pixels
+
+"
+
 # ╔═╡ 3dfccb9a-8f21-4695-965c-c1d97077ba52
 md"# 3. Optimize"
 
@@ -138,12 +146,12 @@ md"Different projection patterns: $(@bind angle PlutoUI.Slider(axes(patterns_via
 simshow(Array(patterns_vial[:,angle,:])[end:-1:begin, :]', cmap=:turbo, set_one=true)
 
 # ╔═╡ Cell order:
-# ╠═c973ccfd-11b7-4e0a-9d1a-32b0d449bfd1
+# ╟─c973ccfd-11b7-4e0a-9d1a-32b0d449bfd1
 # ╠═32d9b7b4-bacd-48c5-8ac5-a7d16bc26f3f
 # ╠═fb00f77e-3a43-43c6-83f5-ab9ffa690585
 # ╠═2a2c5c57-3b51-463b-90b3-bf4733df1c71
 # ╠═9122b9a9-43f4-48be-88f6-6d02a110f900
-# ╠═5c3dda08-010e-4d76-bf35-1c3385a2a0c0
+# ╟─5c3dda08-010e-4d76-bf35-1c3385a2a0c0
 # ╠═efc9b013-0452-482f-9112-d444d2197ba9
 # ╠═54e9b00d-2589-430e-9ac9-ffdb48a54c0e
 # ╠═93591474-679b-4944-a673-616cb5f354e0
@@ -152,14 +160,15 @@ simshow(Array(patterns_vial[:,angle,:])[end:-1:begin, :]', cmap=:turbo, set_one=
 # ╠═5e2ffa51-c23b-4223-89f8-dfa138cfaca5
 # ╠═9aa5e2c5-f0d0-43dc-9c3a-b4cebc5bb100
 # ╠═6869714a-040b-4694-b099-8962445daa33
-# ╠═e7715af7-de78-4668-9733-487f482ac3ef
+# ╟─e7715af7-de78-4668-9733-487f482ac3ef
 # ╠═018120f1-8859-4606-ac62-507fbcaf4097
-# ╠═f1cbc5fd-8454-430c-9267-73c67eca623b
+# ╟─f1cbc5fd-8454-430c-9267-73c67eca623b
 # ╠═3f00e0e5-2d87-406c-9e4a-42b76c3bf2ec
 # ╠═e5bec833-47c5-4f65-8667-f28d4ccb1fbc
 # ╠═e45eeeed-a86e-40e1-b4a2-e4c25cc6a368
 # ╠═f618cc32-6851-42de-b2c2-0ddc70b4668d
-# ╠═3dfccb9a-8f21-4695-965c-c1d97077ba52
+# ╟─29bbf20e-da89-42af-b7a0-a921c1c5fe3f
+# ╟─3dfccb9a-8f21-4695-965c-c1d97077ba52
 # ╠═2350dd2f-11dd-4b81-93b0-a05e88dd0fa7
 # ╠═6eb9c0bc-2aa7-470d-96ca-3716c3c75e4e
 # ╟─819bb337-ef70-4088-82da-bda208264b30