How to Try Meshy 7 Image to 3D - First Indie Alignment Smoke 2026

Meshy 7 went live on August 10, 2026, and its August 12 announcement put image-to-3D alignment—not just a prettier render—at the center of the release. That makes this a useful week for an indie team to test one ordinary prop. It does not make the launch numbers a reason to convert an entire asset list.

Meshy reports that its vendor-designed comparison against four competing models scored Meshy 7 at 81.0% for overall proportion, 79.7% for spatial distribution, and 59.8% for surface details. The reported surface-detail lead is about 5.3 points. The announcement also says single-view input is where systems separate, while the four-view field converges. Those are interesting signals, but they are not independent validation: Meshy designed the benchmark and says it will release the benchmark separately. Read the dated Meshy 7 release announcement beside your own engine screenshots.

This guide turns that launch claim into a small keep/hold decision. You will choose one readable prop, generate from a single reference, compare photo and mesh with six plain-language checks, pass the Pro download gate, then import the result into Unity, Godot, or Unreal. Allow 90–150 minutes. A beginner can complete the viewer comparison even if download is locked; a production decision requires an exported model and an engine smoke.

Open the official Image to 3D Help Center guide while you work. It may still call Meshy 6 the “latest” in places. Treat the Model Version dropdown in your workspace as the operational check: select Meshy 7 if it is available and record what your seat actually shows. Do not rewrite a Help Center lag into a claim that you tested Meshy 7.

Who this Meshy 7 smoke is for

Time estimate: 15 minutes to prepare the reference and asset card, 20–45 minutes to configure and generate, 15–25 minutes for the alignment squint, and 30–60 minutes for export, engine import, and notes.

Honest limits: downloading generated models requires Pro or above. Ultra Mode is currently a single-view path; the release announcement says multi-view is arriving shortly. Availability, labels, and documentation can move, so confirm the model dropdown and plan gate in your own workspace. A viewer result is not an import result. A vendor benchmark is not your art-direction benchmark. One successful prop is not evidence that every prop class will work.

This article does not repeat the Agent, 8K, or tool-list paths

The intent here is narrow: Meshy-7-specific image-to-3D alignment for an indie prop.

• The Meshy 3D Agent evening starts with conversation and concept generation. This smoke starts with a fixed image and asks whether the mesh stays aligned with it.
• The Meshy 8K Texture evening asks whether high-resolution texturing earns its cost on a hero asset. This smoke evaluates geometry, part placement, and relief before texture ambition.
• The AI 3D model generator import-ready checklist compares a broader tool category. This page keeps one vendor, one model version, one prop, and one keep/hold receipt.

Do not combine all three experiments into one night. If the reference itself is still being invented in chat, you cannot tell whether a bad result came from concept drift or image-to-3D alignment. If you jump directly to 8K textures, surface color can hide weak geometry. If you compare five vendors, you will spend the evening normalizing settings instead of learning whether Meshy 7 fits your own pipeline.

Define the keep/hold question before opening Meshy

Write an asset card that can be judged without taste arguments:

Prop: compact desk radio with one large tuning dial, one small volume dial, a rectangular speaker grille, and four short feet.
Reference: one clean three-quarter image with the whole prop visible.
Camera use: waist-high room prop, sometimes viewed from two meters.
Keep if: body proportions, dial positions, grille area, and feet survive the A1–A6 smoke and the engine import.
Hold if: major parts move, disappear, merge, or require a rebuild larger than the time saved.

Choose a prop with a few countable parts and recognizable proportions. A mug with a handle, desk fan, camera, lantern, toolbox, or radio is easier to judge than smoke, tangled foliage, transparent glass, or an ornate creature. The goal is not to find an input that embarrasses a model. It is to pick the kind of prop your game might actually need and define which geometry matters.

Keep the scope to one reference and one primary candidate. You may make a second candidate if the first clearly fails because of an input problem, but write that exception down. Unbounded rerolls turn a model evaluation into survivor bias: the final screenshot looks good while nobody records how many misses it took.

Prepare a useful single-view reference

Use a reference where:

1. the whole object is inside the frame;
2. the silhouette is distinct from the background;
3. important parts are not hidden by hands, props, or text;
4. the viewing angle exposes depth rather than presenting a perfectly flat front;
5. one object is clearly the subject; and
6. the image is the exact reference you are allowed to use in your production context.

These are test controls, not promises about the model. Save the untouched source as reference.png. Do not crop, paint over, or replace it after generation without recording a new run. Your worksheet depends on comparing the candidate to the same source the model received.

Beginner Monday evening path - reference to engine

This path follows one decision chain: workspace → Meshy 7 version → Smart Topology → generate → alignment squint → Pro download → engine import.

Step 1 - Open the workspace and record what it shows

Sign in, open Image to 3D, and make a short note containing:

• date and local time;
• account plan;
• visible Model Version value;
• whether Meshy 7 appears in the dropdown;
• whether Ultra Mode appears;
• whether single-view or multi-view is selected; and
• expected download status.

The Help Center can lag the release announcement. If the page says Meshy 6 while your dropdown offers Meshy 7, record both facts without “correcting” either source. If Meshy 7 does not appear, stop this specific smoke. You may inspect the older workflow, but you cannot label the receipt Meshy 7.

Step 2 - Upload one reference and confirm the mode

Upload reference.png. For this first smoke, stay on the single-view path. The release announcement says Ultra Mode currently supports single-view and that multi-view is coming shortly. That makes single-view the cleanest launch-week test and the place where the vendor says systems separate.

Do not create improvised rear and side views from guesses merely to populate a multi-view field. If you later run multi-view, use genuinely consistent views of the same object and treat it as a separate receipt. Multi-view is not “more images always equals better”; inconsistent shape, scale, or part placement across views creates an ambiguous instruction.

Step 3 - Confirm Meshy 7 in Model Version

Open Model Version and choose Meshy 7 if your workspace offers it. Capture a screenshot or type the exact selected label into your notes. This is the version proof.

If the control still shows Meshy 6 as the latest option, hold the Meshy 7 evaluation. Do not infer that an unlabeled “latest” silently means 7. The distinction matters because the entire article tests claims attached to a specific release.

Step 4 - Choose Smart Topology or Standard deliberately

The Help flow offers Smart Topology and Standard. Treat this choice as a pipeline decision:

• Choose Smart Topology when your smoke includes a usable topology result and you want the generated candidate prepared for a more production-minded inspection.
• Choose Standard when you want the simpler generation path or when your immediate question is only whether the reference’s shape and parts align.

Do not compare a Smart Topology run with a Standard run and call the difference a pure Meshy 7 alignment comparison. Topology choice becomes another variable. For the requested Monday path, select Smart Topology, record it, and keep it constant through the candidate you import.

Topology does not rescue a missing handle, shifted dial, or wrong body ratio. Judge alignment first. Then judge whether topology is workable. A clean edge flow around the wrong shape still fails the prop brief.

Step 5 - Generate one primary candidate

Start generation and name the run before you become attached to it:

radio_meshy7_single_smart_run01

When the viewer opens, resist the beauty-spin. Do not orbit for ten minutes looking for the one flattering angle. Put the mesh near the source image’s camera angle first. Then perform the alignment squint:

1. reduce both reference and viewer to a similar on-screen size;
2. compare outer width, height, and depth impression;
3. count the required parts;
4. compare where those parts sit;
5. inspect silhouette-breaking relief;
6. rotate to reveal whether apparently correct front details are real geometry or only surface appearance.

The first answer is not “beautiful” or “bad.” It is “A1–A6 pass, fail, or uncertain.”

Step 6 - Complete the alignment worksheet before rerolling

Fill every row in the worksheet below. Use uncertain when the reference does not reveal enough information. Uncertainty is not failure, but it prevents an unsupported pass.

If A1, A2, or A3 fails, hold the candidate before spending time on polish. Those gates cover the object’s identity. If only fine relief fails, you can decide whether the prop’s camera distance makes that defect important.

Step 7 - Cross the Pro download gate honestly

Meshy requires Pro or above to download generated models. If your plan is below that gate, the beginner viewer test can still teach you about alignment, but the production smoke stops here. Record:

download_gate: blocked_below_pro

Do not claim “engine-ready” from a browser viewer. Do not ask a teammate to export on an undisclosed seat and erase the plan dependency from the receipt. The plan gate belongs in cost and reproducibility notes.

If download is available, choose the format that your existing engine pipeline already handles. The official Meshy engine integration guide covers bringing Meshy assets into Unity, Unreal, Roblox, and Godot. This article narrows that broad guidance to first-import gates rather than pretending one export setting fits every project.

Step 8 - Import into one real engine project

Use a blank test scene in the engine your game actually uses. Keep a known-size primitive beside the prop, add ordinary lighting, and inspect from the game camera. Do not postpone the test to a DCC-only turntable. Import behavior, scale, materials, normals, pivot, and performance belong to the actual destination.

Finish by writing keep, hold, or rewrite:

• Keep — alignment and import pass for this prop class; allow a second controlled test.
• Hold — promising viewer result, but download, license, import, or a major alignment gate remains unresolved.
• Rewrite — wrong input class or pipeline for the asset; use manual modeling, a different source, or another method.

Meshy 7 alignment smoke worksheet - A1 to A6

The worksheet separates three ideas that often get collapsed into “looks like the photo”: proportion, parts and spatial distribution, and surface relief.

A1 and A2 - judge proportion before detail

Start with the big box. If a short wide radio becomes tall and narrow, detailed grille holes do not save it. Compare the body’s occupied rectangle and the ratio between its largest masses. You do not need a scientific photogrammetry setup; you need a consistent matched view and a decision tied to gameplay.

For silhouette, hide color mentally. A handle that disappears into the body, feet that fuse into a base, or an antenna that moves to the wrong side changes recognition and animation or placement assumptions. Mark each required silhouette feature on the source before looking at the output. This prevents a polished texture from redirecting your attention.

A3 and A4 - count parts, then locate them

Part count answers “what exists.” Spatial distribution answers “where it exists.” A candidate can have two dials and still fail because both drift into the center or swap size hierarchy. Annotate the reference with simple nouns—large dial, small dial, grille, foot one through four—and check them in order.

This distinction also keeps the vendor metrics in perspective. Meshy reports 81.0% overall proportion and 79.7% spatial distribution in its comparison. Those numbers are vendor-reported aggregate scores, not the pass thresholds for your radio. Your threshold is the brief: if moving the tuning dial breaks interaction or the grille occupies the wrong half, the prop fails even if it contributes to a strong aggregate benchmark.

A5 - separate relief from painted suggestion

Rotate the candidate and change the light direction. A dark line that resembles a seam from one angle may be texture or shading rather than a recessed joint. Decide which relief must exist as geometry at the expected camera distance.

Meshy reports 59.8% for surface details, about a 5.3-point lead over the next result in its four-competitor comparison. Treat that as a reason to test, not a guarantee. Surface detail is also the lowest of the three published Meshy 7 figures, so a practical smoke should explicitly inspect relief instead of assuming the headline improvement resolves every groove or grille.

A6 - inspect what one image cannot prove

A single reference cannot fully specify the rear and underside. The model must produce something there, but you should not pretend that inferred geometry “matches” an unseen truth. A6 asks a different question: is the inferred side usable for this game?

Orbit behind the object. Put it against a wall and on the floor in-engine. Look for collapsed backs, stretched connections, open-looking gaps, or a base that prevents placement. Mark uncertain details as inferred. If the game never exposes the rear, a simple coherent back may pass. If the player rotates the prop in an inventory viewer, the same result may hold.

Record the result without hiding misses

Use this compact receipt:

{
  "asset": "desk_radio",
  "date": "2026-08-14",
  "model_version_visible": "Meshy 7",
  "input_mode": "single-view",
  "topology_mode": "Smart Topology",
  "plan_download_gate": "Pro or above",
  "alignment": {
    "A1_overall_box": "pass",
    "A2_silhouette": "pass",
    "A3_parts": "hold",
    "A4_spatial_distribution": "pass",
    "A5_surface_relief": "uncertain",
    "A6_hidden_side_sanity": "pass"
  },
  "engine": "replace-with-unity-godot-or-unreal",
  "decision": "hold",
  "notes": "Replace with observed facts only"
}

The values are an example schema, not a claimed test result. Replace every field with what your run proves. Save matched screenshots beside it so another teammate can audit the decision.

Single-view now and multi-view later

The launch announcement says Ultra Mode is currently single-view, with multi-view arriving shortly. It also says the single-view field is where systems separate while four-view results converge. For an indie team, that supports two distinct tests rather than a rush to one “best” mode.

Test one now: one photograph, Meshy 7 visible, Smart Topology fixed, A1–A6, one engine import.

Test two later: genuinely consistent views of the same physical or designed object, the same A1–A6 worksheet, and a new receipt. Compare whether hidden-side sanity improves enough to justify the extra reference preparation.

For multi-view, keep these controls:

• every image must depict the same object state;
• do not mix open and closed lids, attached and detached parts, or different dial positions;
• label the intended direction of each view in your notes;
• keep the object dominant in every frame;
• reject a set where one image introduces another object as a competing subject; and
• preserve the single-view receipt rather than overwriting it.

These controls do not promise better output. They make your comparison understandable. “Multi-view felt nicer” is not a production result; “A6 changed from fail to pass, while A1–A5 stayed stable” is.

Engine import gates for Unity, Godot, and Unreal

Meshy’s official integration article is the authority for the current import route. Engine versions and project plugins differ, so the gates below describe what to prove, not an invented universal click sequence.

Shared gates E1–E7

Unity first-import smoke

Create a small test scene, import using the current route in the official integration guide, and place the candidate beside a known-size cube. Use a neutral material and ordinary scene light if the imported material obscures the geometry.

Check E1–E7 in order. Do not begin texture tuning before scale, orientation, normals, and geometry pass. Capture one Game view at the prop’s expected camera distance and one close orbit for A5/A6. The Unity gate passes when the model behaves as a placeable prop, not merely when a Project window thumbnail appears.

Godot first-import smoke

Bring the exported asset into a disposable Godot test project using the current Meshy integration guidance. Instance it in a simple 3D scene with a known-size reference and a light. Inspect both the imported scene structure and the visible result.

The practical questions remain the same: can you place it at the origin, does it face correctly, are important surfaces visible, and does its complexity match its role? Take the game-camera screenshot from the expected distance. Godot support in the integration article means there is a documented route; it does not remove your obligation to prove the route with your project configuration.

Unreal first-import smoke

Use a clean map and the current official import guidance. Place the asset beside a known-size reference, light it, and check the front, rear, base, and expected gameplay angle.

Keep advanced optimization decisions out of this first pass. The initial Unreal receipt should answer whether the geometry, scale, orientation, materials, and alignment are usable. If E1–E7 pass, topology and performance work can continue under your normal project standards. If a major part is wrong, engine optimization is premature.

Choose one engine for the decision, not three for show

You do not need to import into Unity, Godot, and Unreal unless your company ships or evaluates all three. Pick the destination engine for your game. The three branches exist so teams can use the same evidence language without pretending the import procedures are identical.

If you are conducting a vendor comparison rather than a Meshy 7 smoke, use the broader import-ready AI 3D tools checklist. That page owns cross-tool normalization; this one owns reference-to-mesh alignment.

License, disclosure, and evidence ownership

The Image to 3D Help material distinguishes Commercial and Personal license contexts. Do not infer rights from the fact that a download button works. At the time of the test:

1. open the license information associated with your account and generated asset;
2. record whether your intended use is commercial or personal;
3. save the relevant plan or license wording with the receipt;
4. identify who approved the asset for the project; and
5. revisit the license check if the plan, account, or intended use changes.

This article cannot decide your rights from outside your account. The safe operational rule is simple: “downloaded” is a technical state; “approved for this game” is a separate documented decision.

Also preserve the source image and its provenance. Record who supplied it, whether it is approved for this test, and where the output may be used. For a company, keep the reference, receipt, screenshots, license note, and exported file under the project’s normal access controls. Do not leave the only evidence in one artist’s personal downloads folder.

Creator checklist and Discord paste

Creators can share a useful Meshy 7 test without turning it into an unqualified “AI replaced modeling” clip.

Creator capture checklist

• Show the exact reference at a readable size.
• Show the workspace’s Meshy 7 selection.
• State single-view and Smart Topology.
• Show a matched reference angle before the dramatic orbit.
• Put A1–A6 on screen, including failures and uncertainty.
• State that benchmark figures are vendor-reported and vendor-designed.
• Mention that the benchmark is expected to be released separately.
• State the Pro-or-above download gate.
• Show one real engine frame, not only the web viewer.
• Name the decision—keep, hold, or rewrite—for this prop class only.
• Keep Agent chat and 8K texture claims out of the alignment result.

Discord paste

Meshy 7 Image to 3D smoke - 2026-08-14
Prop/reference: [name + approved source]
Workspace proof: Model Version showed [exact label]
Mode: single-view | Smart Topology
A1–A6: [pass/hold list]
Download: [Pro+ available or blocked]
Engine/E1–E7: [engine + result]
License context checked: [Commercial/Personal + saved note]
Decision: KEEP / HOLD / REWRITE for [prop class]
Limit: one prop; vendor-designed benchmark is not independent validation
Next: [second controlled prop / wait for multi-view / manual model]

This paste is short enough for a project channel and specific enough to prevent a polished orbit from becoming studio policy.

Company CapEx four-liner

1. Seat: Pro or above is required for generated-model download; buy or retain a seat only when the export-and-engine step is part of the evaluation.
2. Credits: set a written run cap before generation and log actual usage from the workspace; do not estimate unprovided prices in a decision memo.
3. Evidence: require one approved reference, visible Meshy 7 version, A1–A6, exported file, E1–E7 engine screenshots, and license context.
4. Disclosure: record AI-tool use and the owner of any project, platform, client, or partner disclosure decision before a generated asset ships.

That four-liner is deliberately modest. The release metrics can justify a controlled trial. They cannot establish return on investment, legal approval, asset consistency, or production throughput for your studio.

Common mistakes and failure modes

Mistake 1 - Calling “latest” Meshy 7 without dropdown proof

The Help Center may still mention Meshy 6. Capture the Model Version selection. If 7 is absent, hold the 7-specific claim.

Mistake 2 - Treating vendor metrics as an independent benchmark

Quote the numbers as vendor-reported, name the vendor-designed comparison, and note that the benchmark is to be released separately. Your A1–A6 worksheet is a studio smoke, not a replication of the unpublished benchmark.

Mistake 3 - Rerolling until one candidate flatters the announcement

Cap runs. Keep failed screenshots. A repeatable two-run process is more useful than one winner selected from an unknown pile.

Mistake 4 - Letting texture hide geometry

Judge silhouette, parts, placement, and relief under neutral lighting. The separate Meshy 8K Texture guide begins after geometry earns a keep.

Mistake 5 - Confusing Smart Topology with alignment

Smart Topology and Standard are workflow options. Neither changes the identity test: missing, merged, or misplaced parts fail. Record the topology mode so later runs are comparable.

Mistake 6 - Declaring “game-ready” before Pro download

Below Pro, you can assess the viewer but cannot complete the generated-model download and engine smoke. Mark the result as viewer-only.

Mistake 7 - Testing multi-view with inconsistent images

Different object states create a bad control. Save multi-view for a consistent set and a separate receipt, especially while Ultra Mode remains single-view.

Mistake 8 - Inspecting only the source angle

The source angle is essential for alignment, but A6 requires an orbit and engine placement. One-view success can hide a collapsed rear or unusable base.

Mistake 9 - Importing without a known-size reference

An asset can look correct in empty space while being unusably scaled. Put a known-size primitive beside it immediately.

Mistake 10 - Assuming the material preview proves surface relief

Rotate the light. If a seam disappears or remains flat at an oblique angle, it may not be geometry. Decide whether that matters at the real camera distance.

Mistake 11 - Ignoring Commercial versus Personal context

Download access is not a rights memo. Save the applicable license context and assign an approval owner.

Mistake 12 - Generalizing from one prop

A radio keep supports another controlled hard-surface prop test. It does not approve characters, foliage, transparent objects, an entire environment kit, or every camera distance.

Keep, hold, or rewrite decision table

The decision is about fit, not whether Meshy 7 is universally “good.” Keep and hold can both be successful outcomes because they prevent uncontrolled spend and unsupported production claims.

Key takeaways

• Meshy 7 went live August 10, 2026; the August 12 announcement centers image-to-3D alignment.
• Meshy reports 81.0% overall proportion, 79.7% spatial distribution, and 59.8% surface details against four competing models, with an approximately 5.3-point surface-detail lead.
• Those figures come from a vendor-designed benchmark that Meshy says will be released separately; label them vendor-reported.
• The Help Center may still name Meshy 6 as latest in places, so prove Meshy 7 through the Model Version dropdown in your workspace.
• Start with one single-view prop, one approved reference, Smart Topology held constant, and a written generation cap.
• A1–A6 separate overall shape, silhouette, parts, placement, relief, and inferred hidden-side usability.
• Generated-model downloads require Pro or above; a below-Pro viewer check is not an engine-ready result.
• Ultra Mode is currently single-view, while multi-view is described as arriving shortly; keep those tests in separate receipts.
• Unity, Godot, and Unreal decisions need in-engine proof for file import, scale, orientation, pivot, normals, materials, geometry budget, and preserved alignment.
• Check Commercial versus Personal license context separately from technical download access.
• This alignment smoke does not replace Meshy 3D Agent concept work, Meshy 8K texturing, or a multi-vendor import checklist.
• The safest Monday outcome is a scoped keep, hold, or rewrite backed by screenshots and a short receipt.

FAQ

What is Meshy 7?

Meshy 7 is the image-to-3D foundation model Meshy says went live on August 10, 2026. Its August 12 release announcement emphasizes alignment between an input image and generated 3D geometry. For a valid test, confirm “Meshy 7” in your workspace’s Model Version dropdown rather than relying only on a Help Center label.

What do Meshy 7’s alignment scores mean?

Meshy reports 81.0% overall proportion, 79.7% spatial distribution, and 59.8% surface details in a comparison with four competing models. Meshy designed the benchmark and says it will release it separately, so treat the figures as vendor-reported launch evidence—not independent proof that your prop will pass.

Why test Meshy 7 with a single image first?

The release announcement says single-view is where systems separate, while the four-view field converges. Ultra Mode currently supports single-view, with multi-view arriving shortly. A one-image smoke also gives an indie team a controlled starting point: one source, one candidate, one A1–A6 worksheet, and fewer conflicting inputs.

Does Meshy’s Help Center always show Meshy 7 as latest?

Not necessarily. The Image to 3D Help Center may still list Meshy 6 as “latest” in places. Check the Model Version dropdown in your actual workspace. If Meshy 7 is unavailable, do not label an older or ambiguous run as a Meshy 7 result.

Do I need Meshy Pro to try Image to 3D?

You can evaluate what your available workspace permits, but downloading generated models requires Pro or above. Without download access, you cannot finish this article’s engine-import gate. Record the result as viewer-only and decide whether a controlled Pro-seat export test is worth funding.

Should I choose Smart Topology or Standard?

Choose deliberately and record it. Smart Topology fits this production-minded smoke because topology is part of the later engine inspection. Standard can isolate a simpler generation path. Do not change between runs and describe the result as a pure alignment comparison, because topology mode would be another variable.

How do I judge image-to-3D alignment without special software?

Put the reference and mesh at similar sizes and angles. Check A1 overall box, A2 silhouette, A3 required parts, A4 part placement, A5 meaningful relief, and A6 hidden-side usability. Save side-by-side screenshots. A basic image annotation tool and your engine are enough for a first keep/hold.

Can I use multi-view instead of single-view?

Use multi-view when it is available to you and when you have consistent views of the same object. Keep it as a separate test. The launch announcement says Ultra Mode is currently single-view and multi-view is arriving shortly. Do not fabricate extra views or mix different object states simply to fill inputs.

Can Meshy 7 models go into Unity, Godot, and Unreal?

Meshy publishes an official integration guide covering Unity, Unreal, Roblox, and Godot. Your production answer still depends on a real import. Prove file opening, scale, orientation, pivot, normals, materials, geometry budget, and A1–A6 alignment inside the destination engine your project uses.

Does a successful download mean I can use the model commercially?

No automatic conclusion follows from download alone. Meshy’s Help material distinguishes Commercial and Personal license contexts. Check the license information tied to your account, plan, asset, and intended use; save the applicable wording; and assign an approval owner before shipping.

Is this the same as the Meshy 3D Agent or 8K Texture workflow?

No. The Agent path begins with conversation and concept creation. The 8K path asks whether high-resolution texture detail earns its cost. This Meshy 7 smoke begins with a fixed image and measures geometry alignment. Keeping the experiments separate makes failures and costs attributable.

What should an indie team do on Monday?

Pick one countable prop, approve one reference, cap generation, confirm Meshy 7 in the dropdown, select Smart Topology, and run A1–A6. If Pro download is available, import into your real engine and run E1–E7. Post the Discord receipt, record license context, then choose keep, hold, or rewrite.

What to do Monday

Block two hours, not a sprint. Use one prop your game could plausibly ship and one reference your team is allowed to test. Confirm Meshy 7 in the workspace before generating. Hold Smart Topology constant. Complete A1–A6 before rerolling, and stop if the major identity gates fail twice.

If your seat can download, move directly into one destination engine and finish E1–E7. If it cannot, write “viewer-only, blocked below Pro” rather than upgrading on momentum. Save the reference, matched screenshots, receipt, license context, and creator paste in the project folder.

Then make the smallest defensible decision. Keep Meshy 7 for one more controlled prop, hold until multi-view or download access resolves the open question, or rewrite the asset through another method. That is a useful alignment smoke: not a launch-day cheer, not a benchmark dismissal, and not a promise that one impressive orbit has replaced your art pipeline.