diff --git a/README.md b/README.md
index 8560961..1f09112 100644
--- a/README.md
+++ b/README.md
@@ -6,6 +6,28 @@ Explore camera-based inspection of production processes using reproducible illum
 
 **Status:** documentation and source placeholder only. No capture service, trained model, benchmark, validated detection accuracy or deployed application exists in this repository. Hardware has not been selected or purchased as part of this task.
 
+## How it works
+
+The illustrations below show alternative capture arrangements and a proposed operator interface. They are AI-generated concept illustrations, not installation drawings, real defect evidence or screenshots of implemented software. Geometry and defect appearance are illustrative and not to scale. Teal cones represent camera fields of view; amber rays represent illumination.
+
+### 1. Complementary views of the same surface
+
+![Two cameras view the same membrane surface: an overhead view with diffuse light and an oblique view with low grazing illumination.](docs/images/complementary-views.png)
+
+Camera A observes the surface approximately perpendicular to the web; camera B adds an oblique view. Diffuse light reveals surface appearance, while grazing light can make raised features visible through highlights and shadows. Both views target the same inspection region. In experiments, switch lighting recipes separately when needed to distinguish their effects; the drawing shows the available sources together.
+
+### 2. An alternative: inspect both surfaces
+
+![One camera and light above the membrane and another camera and light below inspect opposite surfaces across an open span between rollers.](docs/images/both-surfaces.png)
+
+If both surfaces matter, place one camera and a suitable light on each side of an unobstructed span. Each camera inspects its own surface. This is an alternative to the complementary-view arrangement, not a requirement for four cameras. Mechanical clearance, coverage and triggering still need to be established.
+
+### 3. Detect, highlight and review
+
+![Illustrated monitor showing two membrane images, a red suspected-defect region and human review controls.](docs/images/inspection-monitor.png)
+
+The proposed interface displays the latest captures and highlights a suspected defect on the matching image. The operator confirms or rejects the suggestion; the catalogue retains the image evidence and review history. Optional generated descriptions arrive later and do not delay the alarm. The illustrated monitor shows the complementary-view scenario above; it does not demonstrate detection accuracy.
+
 ## Initial use case: requirements from the discussion
 
 | Constraint | Current scope |
diff --git a/docs/images/README.md b/docs/images/README.md
new file mode 100644
index 0000000..a63bedc
--- /dev/null
+++ b/docs/images/README.md
@@ -0,0 +1,15 @@
+# Concept illustration sources
+
+Created on 2026-09-09 with the built-in Imagegen tool. These are conceptual, AI-generated illustrations, not engineering drawings, measured results or application screenshots. Original PNGs are stored alongside this file and referenced with relative paths in the project README.
+
+## complementary-views.png
+
+Use case: scientific-educational. Create a polished editorial technical illustration for the English README of machine-vision-poc. Wide landscape 1536x1024. Off-white background, graphite equipment, teal camera viewing cones, warm amber illumination, restrained red for a surface defect. Attractive precise isometric 3D illustration with subtle shadows and clean technical callouts, ample whitespace, readable small set of English labels. No branding, no invented measurements, no performance metrics. This is a conceptual illustration, not a real installation. Scene: a dark flexible membrane travels horizontally across two support rollers, unobstructed inspection span between rollers. TWO cameras above the SAME upper surface: Camera A directly overhead looking vertically down, Camera B offset obliquely looking at the SAME small raised wrinkle. Each camera is a plausible compact industrial camera with lens aimed at inspection region. One large diffuse panel light above and offset so it doesn't obstruct camera A. One low LED bar beside membrane casting grazing light nearly parallel to its upper surface. Clearly separate translucent teal view cones from amber light rays. Show subtle shadow cast by wrinkle under grazing illumination. Simple directional arrow along membrane. Exact labels only: '01 / COMPLEMENTARY VIEWS', 'Camera A', 'Camera B', 'Diffuse light', 'Grazing light', 'Web travel'. Do not make light beams look like camera lasers. No screen.
+
+## both-surfaces.png
+
+Use case: scientific-educational. Create a polished editorial technical illustration for the English README of machine-vision-poc. Wide landscape 1536x1024. Off-white background, graphite equipment, teal camera viewing cones, warm amber illumination, restrained red for a surface defect. Attractive precise isometric 3D illustration with subtle shadows and clean technical callouts, ample whitespace, readable small set of English labels. No branding, no invented measurements, no performance metrics. This is a conceptual illustration, not a real installation. Scene: TWO cameras inspect opposite surfaces of one horizontal dark flexible membrane stretched between widely spaced support rollers. Oblique cutaway side perspective makes open air above AND BELOW span visible. Camera A above looking down onto TOP surface; Camera B BELOW the membrane looking UP onto UNDERSIDE. Their fields terminate at respective surfaces; no transparent membrane, no rays passing through it. Small diffuse LED bar on each respective side illuminates its own surface, offset from optical paths. Clearly separate teal camera view cones and amber light rays. Membrane has no table or conveyor bed blocking lower camera. Professional clean geometry. Exact labels only: '02 / BOTH SURFACES', 'Top camera', 'Bottom camera', 'Top light', 'Bottom light', 'Web travel'. No screen.
+
+## inspection-monitor.png
+
+Use case: ui-mockup. Create a polished editorial 3D illustration of a desktop PC monitor displaying a conceptual machine-vision inspection interface, for the English README of machine-vision-poc. Wide landscape 1536x1024, off-white background, graphite slim monitor, subtle studio shadows, slight perspective but nearly front-on for legibility. Same visual language as technical diagrams: teal accents, amber details, red only for defect alerts. Monitor fills most of composition. On screen exact header 'MACHINE VISION POC' and small badge 'CONCEPT UI'. Two side-by-side camera image panels labeled 'Camera A' and 'Camera B', each depicting the same dark grey membrane patch with a clear small raised fold in different lighting. Red bounding box tightly around fold, translucent red defect mask inside region. Prominent red status 'SUSPECTED DEFECT'. Small right or bottom review card with heading 'Human review', suggestion 'Surface anomaly', status 'Unreviewed', buttons 'Confirm' and 'Reject'. Small footer 'Capture → Detect → Review → Catalogue'. No confidence percentages, timestamps, physical dimensions, graphs, fake accuracy claims, logos, people, keyboards or unrelated objects. Beautiful finished product illustration, not a wireframe; English words correct and large. No implication of implemented software.
diff --git a/docs/images/both-surfaces.png b/docs/images/both-surfaces.png
new file mode 100644
index 0000000..f6e801b
Binary files /dev/null and b/docs/images/both-surfaces.png differ
diff --git a/docs/images/complementary-views.png b/docs/images/complementary-views.png
new file mode 100644
index 0000000..dc463b2
Binary files /dev/null and b/docs/images/complementary-views.png differ
diff --git a/docs/images/inspection-monitor.png b/docs/images/inspection-monitor.png
new file mode 100644
index 0000000..bcbdd80
Binary files /dev/null and b/docs/images/inspection-monitor.png differ