I am re-writing my GAME 2325 Character Setup course for the Fall. It is a 16-week course with one class per week that includes a lecture, exercises, and projects where students model, create a skeleton, bind meshes, and create various rigs for a 3D character. The course is primarily in Maya, with export to Unity and Unreal. I also mention Blender and some other external rigging tools. Students have a basic familiarity with Maya, modeling, and animation tools. I'd first like to outline the course, then define the exercise for each class period, then design the projects that add up to a fully rigged character that each student creates for themselves. What does a student need to know to rig a basic bi-ped character, perhaps with wings, a tail, or props that can be independently rigged? --- Ran 2 commands, read 2 files, found files Ran 2 commands, read 2 files, found files Eleven competency domains, plus a draft 16-week arc mapped to them and six decisions that will change the outline. Three things worth flagging from building it: - **The unifying idea for the whole course** is that a rig is an interface for producing joint transforms. Controls, constraints, IK, and deformers don't export — only joints and skin weights do. If students absorb that in week 1, the export week stops being a surprise and the "why bake?" question answers itself. - **Wings/tail/props are three different problems.** Tails are a chain-solver problem (FK vs. spline IK vs. ribbon vs. dynamics). Wings are a Set Driven Key problem — one `fold` attribute driving a feather hierarchy. Props are a space-switching problem plus a socket-joint handoff to the engine. Grouping them as "extras" hides that they teach distinct skills. - **Your project sequence falls out naturally in three beats:** model + skeleton → bind + weights + ROM test → full control rig + appendage + engine export. Each is independently gradeable and the third can't be faked without the first two. Your project knowledge folder is currently empty (metadata only), so this is built from scratch — if you have the existing syllabus or prior semester's schedule, upload it and I'll reconcile against it before we lock the outline. Once you settle the six decisions at the bottom, I'll write the week-by-week outline, then the exercises. C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\GAME2325_Competency_Map.md Document --- Student will model their own character. Some have a model from a previous course, while others will create a new model. So, the course needs to cover deformable topology. I have a simple topology puzzle that I introduce to show the key solution to redirecting edge loops and transitioning from two face loops to three face loops. T-post is the preferred standard, so that bones can be drawn on the axis for the spine, legs, and arms. Unity is the primary target for export, though we also need to discuss Unreal export. Both game engines are used in the academic program. I provide a limited number of scripts for renaming (when creating 50 blendShapes for the face, e.g.), but the course does not cover syntax. I touch on facial rigging, but do not require it. I have included Jason Osipa's "Stop Staring" book as a reference, and mention more advanced methods. Many students will have characters with tails, wings, or weapons like a hammer, gun, whip, or bow. Riggins props are part of the character rigging process, so we can talk about space switching. I stress that no rig will cover every situation and that we rig for the shot. A rig is a last step in creating a character before animation, and a character may be re-rigged for a different shot. --- Created a file, read a file Created a file, read a file Four units, three projects, sixteen weeks. Four structural moves I made from your answers — flag any you'd reverse: **Your "rig for the shot" principle became the grading mechanism.** Students write a _shot spec_ in Week 1 alongside their character, and P3 is graded against their own spec. The student with a bow is graded on the bow; the student with wings is graded on wings. It turns a philosophy into a rubric and it solves the problem of grading wildly different characters fairly. **Two parallel tracks throughout.** In-class exercises run on an instructor-provided practice mesh; projects run on the student's own character. Nothing in class waits on someone's model being finished, and the mixed-readiness problem stops being a scheduling problem. Cost: you need to author that mesh plus sabotaged variants for the diagnostic exercises in Weeks 2, 6, and 15. **Appendages and props split into two weeks, not one.** Tails, wings, and whips are all chain-solver problems (Week 13). Weapons are a space-switching and socket problem (Week 14). A bow is genuinely both — chain for the string, two-handed constraint for the grip — which makes it a good demo character if you want one. **Facial landed in Week 12** with head and neck, where the jaw and eye-aim work belongs anyway, and where your blendshape rename script has a natural reason to appear. Osipa as reference, elective for projects. One thing worth your attention: a student with a weak model or skeleton at Week 4 compounds problems for twelve weeks. I'd suggest a mandatory revision gate on P1 rather than just a low grade — noted in the open items along with grade weighting. Next step is the sixteen exercises. Week 2's topology puzzle is the model I'd write the rest against — one key insight, learned by doing rather than watching. Want me to draft all sixteen, or work through them unit by unit so you can course-correct the format early? C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\GAME2325_Course_Outline.md Document --- Shouldn't one project be the character model, with a propose revision for correction, then a project for skeleton and naming, then bind, then the last project be the rig, but broken up into sections: main & hips (for layout), leg and foot rigs, spine rig, arm rigs, neck & jaw. I use a top node called MSR (MoveScaleRotate) and the hierarchy resolves to this one node, with a group each for Geometry (Meshes), Joints, and Controls. Because we might use multiple characters in a scene, referenced into a Sequence_Shot scene, each character, prop, vehicle, or environment ("CPEV") resolves to a single node from a single file. I also need to cover a bit of the production pipeline, which I introduce early on so students can understand how they are focusing on one part of a larger system. --- Created a file, read a file Created a file, read a file Restructured to four projects with P4 broken into your six milestones. CPEV and MSR are now stated up front as the two conventions the course runs on, and the pipeline overview opens Week 1. Judgment calls worth checking: **Week 9 does double duty.** M1 is only three or four controls, so I used that week to teach the full control vocabulary — curves, zero groups, attributes, constraints, utility nodes, matrix workflow — with the MSR/main/hips layout as the exercise that applies it. The two fit because building M1 correctly requires exactly those fundamentals and nothing more. It also puts the MSR uniform-scale problem where it belongs: solved once, at the top, before any IK exists to break. **Props became a CPEV lesson, not just a rigging lesson.** A hammer is its own file with its own top node, referenced into the shot beside the character — not modeled into her hand. That reframing makes space switching the obvious consequence rather than a technique students memorize, and it reinforces the convention a second time in Week 14. **Space switching moved to Week 12** with arms and hands, so students meet it on a hand before they need it on a weapon two weeks later. **Weeks 1–4 spend three sessions on the model** — topology, then file setup with in-class critique, then the revision gate. That's a quarter of the semester before a joint gets drawn on their own character. Defensible given the revision gate depends on it, but it's the tightest budget decision in the outline. If you want a week back, Week 3's critique could move to studio time and skeleton work could start a week earlier. Two things I'd want from you before writing exercises: the actual topology puzzle as you teach it, so the other fifteen match its shape, and whether the practice mesh already exists or needs authoring. C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\GAME2325_Course_Outline.md Document --- Grades should add up to 1000 points, so it is easy to determine a letter grade through a percentage. I have a practice mesh that I have used in the past, but I will develop a new mesh with better topology and proper scaling. The Rig spec should cover everything the character needs to be able to perform in a scene, with the corresponding rig elements determined. This could be a checklist of all the possible rig elements the course develops, with a few requirements and all the possible rigs, like tails, wings, stretchy IK, facial rig, etc. Design drawing can be registered in Photoshop with guides. They have done this in the past with their 2D characters for Storyboard & Concept Development class. They might even be able to use the same character. The Design drawings should show Front, 3/4, side, back, and the other side if the character is asymmetrical. There should be callouts for any special features or items, like props, or wings tucked and spread. The pipeline follows Pixar's pipeline starting with story, concept, design, then moving into 3D with model, texture, rig, lighting, layout, animation, and back to 2D with rendering, editorial, and adding special effects, sound, voice-over (which will come before animation to sync), and final composite. --- Edited 4 files, read a file Edited 4 files, read a file Grading built to 1000, the rig spec is now a full checklist, and the pipeline follows the Pixar three-movement structure. **One thing I changed from what you wrote — confirm or correct me.** You listed the 3D stage as "model, texture, rig, lighting, layout, animation." I ordered it _Model → Texture → Rig → Layout → Animation → Lighting_, since layout blocks cameras and staging before animators work, and shot lighting is normally finalized after animation is locked. I also gave voice-over its own callout as the stage that jumps its apparent position — a useful way to make the point that a pipeline is a dependency graph rather than a straight line. If your version of the sequence is deliberate, say so and I'll restore it. **The rig spec became a three-tier checklist:** seventeen required rows mapped to the milestone that delivers each, four electives where students choose at least one, and fifteen optional items for shot-driven credit. Every elective row demands a stated shot reason. That form is also the Week 16 critique rubric — if the spec says the wings spread, the demo animation has to show them spreading. **Points breakdown:** 450 projects / 300 milestones / 150 final / 100 exercises. Weeks 9–14 have no exercise line because the exercise _is_ the milestone, which keeps the arithmetic honest. Two things worth a second look. **P1 is due Week 2** — concept, registered design sheet, and rig spec in one week. If your students are bringing characters forward from Storyboard & Concept Development that's fine, but a student starting cold has a thin week. **And 125 points on the model** may be light for something that carries a mandatory revision and gates twelve weeks of work; it's the only place I'd consider shifting weight. Ready to write the exercises whenever you are. C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\GAME2325_Course_Outline.md Document --- Week 1: I have used a bouncing ball rig to demonstrate a hierarchy rig: MSR > translate > scale > rotate > sphere mesh. This allows a .squashStretch attribute on the MSR to be connected via Set Driven Key to the scale attributes of the scale node. There are no joints and the mesh is a child of the lowest deformation node, but the idea of a hierarchy is the same. This might be a good exercise. As a bonus, we can rig the ball to roll along the ground without slipping by an expression of the circumference of a circle related to its radius. The MSR is usually placed in layout for each scene. Week 2 should cover the topology of muscles. If you take half a cube (open at the half-way point) and smooth it (or extrude), you get the boundary fitting into the edge loops of the arm, and denser topology in the center for stretching and bulging. You can take a square or rectangular patch of polygons and extrude them to make the muscle topology. Week 3: if the design images, broken up into front, side, back are aligned and exported from Photoshop with the same height, they will line up in the orthographic and perspective views in Maya. Week 4: since Maya has joints and not bones, the joint orientation is determined by the next joint in the chain. The lecture should cover joint orientation, evaluation order, re-orienting joints, not rotating joints but moving them to position them, and even redrawing a skeletal hierarchy if necessary. Week 5: the hands typically have a "hand" joint off the wrist that acts as the orientation joint and an attachment point for props held in the hand. In early games, this was called a "gun" joint. If you have good points on heatmaps and geodesic voxel or some other method, I am willing to explore those. Week 6: For Maya, I use dual quaternion weighting and both the component editor and the Paint Weights tool to adjust the weighting. Week 7: please define ROM when first used. Week 8: You can edit the deformer order in Maya if the blendShape added after the rig seems to break the rig, just make sure it is placed before the skinweight. All else looks good. It's time to add detail: Let's build Exercises, which should be 10-20 steps, unless, like the reverse foot, they need to be a bit longer. Please remember that this may be the first time students are encountering these specific tools, though they should be familiar with Maya. Let's build Lectures. I can fill-in if there are missing points (I have certain key phrases I've been using for 20+ years that I sometimes only remember while I'm giving the lecture!), and build a Maya lecture file for each topic. I also have to build a practice mesh with good topology and canonical skeleton, plus sabotaged variants. I can record demonstration video. Any reference videos you can find for each topic and exercise would be very helpful. --- Edited 6 files, searched the web, loaded tools Edited 6 files, searched the web, loaded tools Unit 1 is built — four lectures, four exercises, reference links per topic. Your Week 5–8 notes are folded into the outline so nothing's lost. **Two format choices worth confirming before I write the other twelve:** **Every exercise has a marked "discovery" step where students break it first.** Week 1 step 20 has them rotate the scale node and watch the squash go diagonal, _then_ learn one-job-per-node. Week 4 step 7 has them move a joint and see the orientation go stale, _then_ get the rule. Your topology puzzle works this way — the unstated side-exit constraint is the whole lesson — so I built the rest to match. It costs a step or two per exercise and it means the rule arrives as an answer to a problem they already have. **I made MSR a NURBS circle rather than an empty group.** It gives you the transform-vs-shape distinction on screen in Week 1 step 6, and it means MSR is selectable in the viewport from the start. Easy to revert if you prefer an empty group. **Your ball rig gained a step I'd defend.** Step 28: scale MSR to 2 and the roll expression slips, because it still assumes a radius of 10. That's the course thesis landing in Week 1 — the rig was built for a shot and breaks outside that shot's assumptions. **On references:** the Rigging Dojo article on joint orient is the strongest thing I found and I'd assign it outright for Week 4. Topology Guides is the right standing reference for Week 2 — its loop-redirection pages are your puzzle in another form. Video coverage for Weeks 1–2 is thin and mostly aimed at animators rather than riggers, which argues for your own recordings there more than for Week 4. One gap I couldn't fill: I found no good video specifically on registering design views for image planes. That may genuinely be a recording worth making. Say the word and I'll do Unit 2 — Weeks 5–8, where your hand/gun joint note, dual quaternion default, and the deformer-order fix all need exercise steps. C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\GAME2325_Unit1_Lectures_and_Exercises.md Document  C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\GAME2325_Course_Outline.md Document --- It looks good so far. Please proceeed! --- Searched the web, created a file, edited a file Searched the web, created a file, edited a file Unit 2 built. One finding you should weigh in on before I go further. **Dual quaternion doesn't survive the trip to the engine.** Game engines skin with linear blend by default, and Maya's DQ skinning doesn't transfer to Unreal — the skinning method has to be set to classic linear on export. So weights that look right in Maya under DQ can deform noticeably worse in Unity. I built Week 6 to keep DQ as your Maya default but added step 21: flip the skinCluster to classic linear, run the same range test, and decide whether the weights survive it. That pulls the Week 15 "what you see in Maya isn't what ships" thread forward to where it's actionable. Tell me if you'd rather handle it differently — it's the one place where your stated preference and the export target pull against each other. **Bind methods got a real comparison table** since you were open to exploring them. Geodesic voxel as course default — it was built specifically for production meshes with layered geometry where heat map fails, which describes most student characters with armor or clothing. Week 6 Part B has students bind three ways and write two sentences comparing them, so the choice is earned rather than told. Chris Evans' write-up on why the method exists is the best thing I found and worth assigning. **The sabotaged weights file needs four failures, and one of them must be topological.** That's the design that makes Week 7 teach diagnosis instead of technique — students who can't tell a weight problem from a topology problem will paint for twenty minutes on a mesh that can't deform. It's the same distinction they made in Week 2 between count failures and direction failures, which is why it's worth seeding deliberately. **Your hand joint got its own lecture segment** with the "gun joint" history intact, and it's flagged as non-deforming with a forward pointer to Week 14. One convention decision I left open because it's genuinely yours: whether finger chains parent to the wrist or to the hand joint. Either works, but it affects Unity's Humanoid mapping and students need to be told which. Unit 3 next — Weeks 9–14, the six milestones. That's the largest block and where the matrix workflow and space switching need careful step-by-step treatment. C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\GAME2325_Unit2_Lectures_and_Exercises.md Document --- Finger joints should parent to the wrist joint. The Hand joint is for orienting the wrist joint, but also functions as a transform attachment point for props. When binding, I select the joints that the mesh will bind to. This means that some joints have a _bnd suffix, while some are for placement (like a L_eye_group) with a _grp or _plc suffix, and others, like the finger tips, have _tip suffix because they are not used in the bind. Once the joints are selected, I'll select the mesh and create a Quick Select Set to contain the select. The Bind Joints command uses "Selected Joints" instead of "Hierarchy" to bind to the specific joints. I have not seen issues with Dual Quaternion weights in Unity. But, I do want to keep the exploration of other bind options. If the above affects the Lectures or exercises, please redo Unit 2. Otherwise, head to Unit 3. --- Thought process Thought process Your bind workflow and suffix scheme change Week 5 and 6 materially. Redoing Unit 2. Edited 3 files, created a file, read a file Edited 3 files, created a file, read a file Unit 2 redone. Your suffix scheme turned out to be load-bearing, so it propagated backward into Unit 1 and the outline too. **The suffix now has a job, and the whole unit hangs on it.** `_bnd` isn't a label, it's the filter that produces the bind selection. I made that explicit in Week 5 — students count their `_bnd` joints, then check that number against the skinCluster's influence list in Week 6 step 8. If it's longer, they bound to the hierarchy by mistake. That turns naming discipline into something with an immediate, checkable consequence instead of a rule they follow because you said so. **The Quick Select Set earns its keep in the same exercise it's taught.** Week 6 Part C has students rebind three times to compare methods — which is tedious hand-picking forty joints and trivial with the set. They feel the payoff twenty minutes after building it. **I added a fifth failure to the sabotaged weights file:** a `_tip` joint that got into the bind. The fix isn't to remove an influence by hand, it's to rebuild the set and rebind — which teaches that a bad bind is cheaper to redo than to repair, and reinforces the same "redrawing is normal" principle you wanted for skeletons in Week 4. **On dual quaternion, I deferred to your experience.** DQ stays the course default and the Maya-vs-engine step came out of the exercise. FBX has carried DQ data since 2012, so your Unity result makes sense. I kept one sentence in the lecture noting Unreal as the known exception, and reframed the general point as a habit rather than a warning: verify in the engine rather than assume it matches Maya. That belongs in the course anyway and it now costs nothing. **Two small propagations you should know about:** Unit 1 Week 4 renaming changed to `L_hip_bnd` / `L_toe_tip` style with uppercase side prefixes, and the Week 1 naming lecture now carries the suffix table with a forward pointer to the Week 6 bind. I picked up `L_` uppercase from your `L_eye_group` example — say if that's wrong. One open question I couldn't resolve from your notes: the hand joint isn't bound, so it shouldn't be `_bnd`, and I've written it as `L_hand_plc`. If you use a distinct suffix for attachment sockets, tell me and I'll change it before Week 14, where every prop attaches to one. Unit 3 next. C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\GAME2325_Unit2_Lectures_and_Exercises.md Document --- Okay, on to Unit 3! --- Writing file… ·~44,474 characters Searched the web, created a file, read a file Searched the web, created a file, read a file Unit 3 done — six weeks, six milestones, 300 points. **The strongest structural thing in this unit is a two-week callback.** Week 9 teaches the MSR scale problem in the abstract, then M1 step 22 asks students to _write a prediction_ about what will break when IK arrives. Week 10 step 33 has them scale MSR and watch the stretchy leg break exactly as predicted, then fix it by dividing the measured distance by global scale. It's the same bug as the ball rig's roll expression in Week 1. Three encounters with one idea, each a level deeper. **Set Driven Key appears three times on purpose** — foot roll (W10), hand poses (W12), wing fold (W14). By the third students should recognize the pattern instantly rather than learning a new technique. Same with space switching: hand in M4, head and eyes in M5, prop in M6. **Week 11 pays off your joint orientation convention.** Spline IK's Advanced Twist Controls assume X is the aim axis — a chain oriented any other way simply won't twist. That's seven weeks of "X down the bone" finally having a concrete consequence, and step 12 has students diagnose a deliberately broken file before they hear why. **Props teach CPEV by failing.** Step 14 asks students to parent a referenced prop under the hand joint. Maya won't let them cleanly, because a referenced node's parent belongs to its own file. That's the moment constraints stop being an arbitrary choice. **Two things needing your call:** **Week 10 is oversized** — forty steps covering three chains, IK/FK blend, reverse foot, and stretch. I flagged it as spilling into homework, but it may want to borrow time from Week 11, which is comparatively light. Splitting stretch into Week 11 would balance them. **I proposed `main_ctrl` as separate from `MSR`** — MSR belongs to layout, main_ctrl belongs to the animator, so a layout artist repositioning a character never touches animation. I also split COG from hips so the pelvis can move while the spine holds. Both are my inference from your MSR description rather than something you specified. If your actual hierarchy is flatter, tell me and I'll collapse it before Unit 4. Unit 4 — Weeks 15 and 16, export and critique — whenever you're ready. C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\GAME2325_Unit3_Lectures_and_Exercises.md Document --- Since IK handles will be parented under the Controls node, they scale just fine with the rest of the rig. The only adjustment is for the geometry node, which is the parent of all bound meshes, where the "Inherit Transforms" property is turned off. Any transform of the meshes is handled by the components bound to the joints, not by the transform nodes. There really is no "MSR scale problem" with this method. In the Week 1 ball rig, the radius part of the equation should contain MSR.scaleX multiplier, as all of the axis of the MSR should be scaled simultaneously. This should compensate the transform according to the radius of the ball. A larger ball has a larger circumference. I saw this, but had to think about whether it mattered. --- Can we write each unit/week/lecture/exercise/project/milestone for a student perspective? I will be publishing these in a Canvas LMS and will need to break each into a separate page with links back to each week. Most of the language is fine for students, but I need to have anything addressed to the instructor removed or rephrased. --- Thought process Thought process Done Before I generate ~30–50 files, one structural fork worth settling. How should the graded work be split from the reading content?Pages + Assignments (Recommended) How should I deliver these?Unit by unit (Recommended) Building Unit 1 as Canvas pages and assignments. Created 6 files Created 6 files Now the two reference pages, then the assignments. Created 2 files Created 2 files Now the six assignments. Created 6 files, read a file Created 6 files, read a file Thirteen files in a `canvas/` folder — eight pages, five assignments, plus the Home page. **Each file opens with an HTML comment block** giving the Canvas page title, suggested slug, type, points, and due timing. Canvas ignores HTML comments on render, so you can paste the whole file and the metadata stays invisible while remaining visible in the source. Links use short slugs — `week-1`, `project-2`, `naming-convention` — so they resolve if you title pages to match. **What I stripped:** lecture timings, demo-file build lists, "demonstrate the failure," "ask the class," "drill this," and the discovery notes that begin "Not stated to students." **The delicate part was the discovery steps.** Those exercises depend on students hitting a wall before hearing the rule, so I couldn't just delete the framing or leave it in. They became plain instructions that produce the experience without spoiling it — Exercise 1 step 20 now reads "Try this, then undo it. Rotate `ball_scale_grp` 45° in Z…" followed by a request to write down what it told them. Exercise 4 steps 6–8 have students rotate a joint, move a joint, then write the two rules themselves. The learning still arrives in the right order. **I had to invent the point breakdowns.** Projects 1, 2, and 3 previously had pass criteria but no weighted rubric, and Canvas assignments need one. Project 2's 125 points now allocate 30 to joint topology and 15 to file structure, for instance. Those weightings are my judgment — worth a look before I apply the same pattern to Projects 4 and 5. **Three pages are linked but not yet written:** Engine Legality Checklist, Vocabulary, and Resources. They belong with Unit 2 and the final batch, so links to them will be dead until then. Two things worth your eye before I do Unit 2: whether the tone is pitched right — I aimed for direct and non-condescending, with the "why" attached to rules rather than assumed — and whether Week 3's ungraded walkthrough belongs on the page or should become a zero-point Canvas assignment so students can check it off. C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\00_PAGE_Home.md Document  C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\01_PAGE_Unit1_Overview.md Document  C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\02_PAGE_Week01.md Document  C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\08_ASSIGNMENT_Exercise01_Bouncing_Ball_Rig.md Document  C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\11_ASSIGNMENT_Project1_Concept_Design_RigSpec.md Document --- Is there an Exercise03? Everything else looks great! --- Thought process Thought process Done No — and that was a numbering gap I should have caught. Exercise numbers match their week, so there's no 3 (Week 3's walkthrough folded into Project 2) and no 9–14 (those weeks' builds are milestones). The 9–14 gap reads as design; the gap at 3 just reads as a mistake. Making Week 3 a real Exercise 3 fixes it and adds something useful — a registration checkpoint before Project 2 is graded. Edited 3 files, created a file Edited 3 files, created a file Now the Home page and the master outline, to keep the 100-point exercise total intact. Edited 2 files, read a file Edited 2 files, read a file Exercise 3 now exists as a 5-point checkpoint, and the numbering scheme is stated on both the Home page and the Unit 1 overview so the 9–14 gap reads as design. **Why a checkpoint rather than a full exercise.** Forty-five minutes, graded on four things. Its real job is catching unregistered design views _before_ students model against them — right now that error surfaces in Week 6 as bad deformation, which is a terrible place to discover your side view was forty pixels off. It also forces the asset file into shape once, at the start, instead of being corrected on every submission afterward. **To hold exercises at 100, Exercise 8 drops from 15 to 10.** The elbow corrective is also assessed inside Project 4's 150 points, so it was the one place with genuine overlap. Category totals and the 1000-point course total are unchanged. I updated the master outline too, so the two documents don't drift. The Week 3 page now summarizes and links rather than duplicating the steps — the walkthrough lives in the assignment where students will actually have it open. Full exercise list, for the record: 1, 2, 3, 4, 5, 6, 7, 8, 15. Weeks 9–14 are milestones instead. Unit 2 next — Weeks 5–8, Exercises 5 through 8, Project 4, plus the Engine Legality Checklist page that Project 3 already links to. C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\14_ASSIGNMENT_Exercise03_Image_Planes_and_Asset_File.md Document  C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\04_PAGE_Week03.md Document  C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\00_PAGE_Home.md Document --- Okay on to Unit 2! --- Created 6 files Created 6 files Created 15_PAGE_Unit2_Overview.md Created 16_PAGE_Week05.md Created 17_PAGE_Week06.md Created 18_PAGE_Week07.md Created 19_PAGE_Week08.md Created 20_PAGE_Engine_Checklist.md Done Now the four exercises and Project 4. Created 5 files, read a file Created 5 files, read a file Eleven files. Unit 2 overview, four week pages, the Engine Legality Checklist, four exercises, and Project 4. **Three things worth your eye:** **Exercise 7 tells students up front there are five failures and that one can't be painted out.** In your version that was the hidden discovery. I kept it hidden in Exercise 6 and 8, but here I revealed it — because Part B asks them to diagnose all five _in writing before fixing anything_, and that instruction only makes sense if they know how many they're hunting. The difficulty stays intact: knowing one is topological doesn't tell them which one. Four of the fifteen points ride on diagnosing before touching, which is the habit worth grading. **Project 4's rubric puts 25 points on elbows and knees** — the single largest line — because that's where the three-loop rule either paid off or didn't. Bind construction gets 20, which may look high for something that takes five minutes, but it's the step that determines whether the influence budget is holdable at all. Adjust if you'd weight it differently. **Project 4 has a "note on scope" at the end** telling students that correctly identifying a topology problem earns more credit than an hour of painting that can't work. That's the unit's real lesson, and it needed saying explicitly or students will assume admitting defeat costs them points. The Engine Legality Checklist is now live, so Project 3's link resolves. It's written to be used three times — Project 3, Project 4, and export — with the export section marked as not-yet-needed so it doesn't alarm anyone in Week 5. Unit 3 next: Weeks 9–14, the six milestones, and the Project 5 umbrella page. C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\15_PAGE_Unit2_Overview.md Document  C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\20_PAGE_Engine_Checklist.md Document  C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\23_ASSIGNMENT_Exercise07_Repair_and_ROM.md Document  C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\25_ASSIGNMENT_Project4_Bind_and_Deform.md Document --- proceed. --- Created 7 files Created 7 files Created 26_PAGE_Unit3_Overview.md Created 27_PAGE_Week09.md Created 28_PAGE_Week10.md Created 29_PAGE_Week11.md Created 30_PAGE_Week12.md Created 31_PAGE_Week13.md Created 32_PAGE_Week14.md Done Now Project 5 and the six milestone assignments. Created 7 files, read a file Created 7 files, read a file Fourteen files. Unit 3 overview, six week pages, the Project 5 umbrella, and six milestone assignments. **Project 5 is a Page, not an Assignment.** There's no single submission for it — the six milestones are the graded objects and they sum to 300. Making it an assignment would create an empty gradebook column. It carries the shared standard instead: everything inside `MSR`, every control zeroed and in a zero group, colored by side, unused channels locked. Those aren't repeated in each milestone rubric, so students should be told once that they're always in force. **Three things worth checking:** **M2 is the outlier at 42 steps** and I flagged it in the assignment itself as expected to run past class. Its rubric splits 55 points across nine criteria, with 10 on the blend and 8 each on the pole vector, the reverse chain, and the foot roll. If you'd rather move stretch to Week 11 — which is comparatively light — Parts G and the last three points lift out cleanly. **M6 has branching requirements.** Parts A–D are identical for everyone; Part E splits three ways by rig spec, including a fallback for characters with no appendage at all (build a second prop with a swap). The rubric handles this with 15 points on "appendage built per your rig spec" plus a short additional-criteria section per type, so the same assignment grades a wing and a tail fairly. **The referencing lesson stayed a do-it-yourself moment.** M6 step 14 asks students to try parenting a referenced prop under the hand joint, observe what Maya does, and write a sentence about what that means for a shot with six props. That's the CPEV convention paying off by resisting them rather than by being explained again. Last batch is Unit 4 — Weeks 15 and 16, Exercise 15, the final demonstration — plus the Vocabulary and Resources pages that have been linked since the Home page. C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\26_PAGE_Unit3_Overview.md Document  C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\33_ASSIGNMENT_Project5_The_Rig.md Document  C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\35_ASSIGNMENT_Milestone_M2.md Document  C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\39_ASSIGNMENT_Milestone_M6.md Document --- Finish it up! --- Created 7 files Created 7 files Now the two reference pages that have been linked since the Home page. Created 3 files, ran a command, read a file Created 3 files, ran a command, read a file Done. Forty-nine Canvas files plus eight instructor documents. **I ran a link audit across all 49 files** — every slug referenced is defined somewhere, and every slug defined is referenced somewhere. No dead links, no orphan pages. **The final 150 split into three assignments** rather than one, because they're graded three different ways: Export Verification is a file check, the Demonstration is media, and Critique is live participation. One 150-point column would have hidden that, and students would have no way to see which part they lost points on. **A judgment call on the Demonstration rubric:** 25 of its 75 points are spec coverage — the largest single line. That makes the Week 1 rig spec worth more at the end of the semester than the beginning, which is the point. It also means a student who scoped carefully and built everything they claimed beats one who claimed more and delivered less, even if the second rig is technically fancier. Both assignments say so explicitly, because students otherwise assume ambition is always rewarded. **Critique has a rubric line for participating in others' critiques**, with examples of useless versus useful feedback. Without that, step 3 and step 4 of the protocol collapse into "looks good" and the format stops working. **What you have now:** ||| |---|---| |Instructor|Competency map, course outline, four unit guides with lectures and full exercise steps, source list with gap analysis, vocabulary master| |Canvas|25 pages, 24 assignments — home, four unit overviews, sixteen week pages, four reference pages, nine exercises, five projects, six milestones, three finals| What's left is the part only you can do: working the exercises to verify the step counts and menu paths against your Maya version, and building the roughly forty demo and sabotaged files listed in the asset tables at the end of each unit guide. The practice mesh with its known-good skeleton is the long pole — Weeks 2, 6, 7, and 15 all need broken variants of it. C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\40_PAGE_Unit4_Overview.md Document  C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\45_ASSIGNMENT_Final_Demonstration.md Document  C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\46_ASSIGNMENT_Final_Critique.md Document  C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\47_PAGE_Vocabulary.md Document  C:\Users\jmars\AppData\Roaming\Claude\local-agent-mode-sessions\9aa07aca-8786-4e12-8058-397700013b46\e27a738f-052a-46d9-8abf-9ab6e4b14bb8\local_95a3c9e9-d427-457c-9503-ea3112a279a1\outputs\canvas\48_PAGE_Resources.md Document --- ``` This schedule is great for a regular semester. This Fall, however, this course has only 15 classes, with Week 14 being Thanksgiving Break. Can we adjust the schedule for this exception? ``` --- Thought process Thought process Done Week 14 dark means 15 sessions for 16 sessions' worth of content. Something has to merge, and there are two clean places to do it.