Migrate from Face Skin Metrics 2.0 to Face Skin Analysis 3.0

What changed between Face Skin Metrics 2.0 and Face Skin Analysis 3.0, and how to migrate your integration to the new version.

Face Skin Metrics 2.0 will be deprecated by the end of 2026. Move your integration to Face Skin Analysis 3.0 before then. This page is the migration guide.

Starting a brand-new integration? You do not need this page. Follow Configure Your App Backend and SaaS for Integration and Integrate Haut.AI Image Analysis with Your App.

What changes

If you upload with signed URLs, the photo upload itself stays the same: you still request a signed URL, PUT the file to it, and then send the batch to computation. What changes is the endpoint versions, how results are read, and how results are shaped.

If you upload photos as base64 (b64data) to the v1 images endpoint, the analysis starts on upload and there is no compute call. Switch to the signed URL upload and the compute call below.

	Face Skin Metrics 2.0 (old)	Face Skin Analysis 3.0 (new)
Application tech_name	face_skin_metrics_2.0	face_skin_metrics_3
Request an upload URL	POST /api/v1/.../subjects/{subject_id}/signed_upload_url/	POST /api/v3/.../subjects/{subject_id}/upload/
Upload the file	PUT to the signed URL	Same. PUT to the signed URL
Base64 upload	b64data in the v1 image upload, which starts the analysis on upload	Not used. Request a signed URL, PUT the file, then call compute
Group of photos	image_batch_id from the upload response, passed as batch_id	Same batch, passed as image_batch_id
Start the analysis	POST /api/v1/.../batches/{batch_id}/compute/	POST /api/v3/.../batches/{image_batch_id}/compute/
Read results	GET /api/v1/.../images/{image_id}/metrics_and_submetrics/, once per image	GET /api/v4/.../batches/{image_batch_id}/results/, once per batch
Result shape	An array of metric objects	One face_skin_analysis_3 object. Each parameter is a key, for example face_skin_analysis_3.parameters.pores
Masks and aux images	GET /api/v1/.../images/{image_id}/aux_images/ and the metrics response	Inside the same results response: parameters that support masks carry them, and predicted_images holds the images
Results history for a subject	GET /api/v1/.../subjects/{subject_id}/metrics_and_submetrics/	GET /api/v4/.../subjects/{subject_id}/results/
Photo quality	Image quality algorithm inside the 2.0 result	parameters.quality (front, left, right)
Webhook payload	Webhook v1 payload with the 2.0 metrics	The Webhook v1 batch payload also carries the 3.0 results, under face_skin_metrics_3. The per-photo v1 payload does not. Webhook v3 is recommended

Migration steps

1. Attach Face Skin Analysis 3.0 to your dataset

2. How to upload

3. How to send to computation

4. How to get aux images

5. How to get results

6. How to get and show masks

7. How to know when results are ready

8. How to map old results

9. How to turn off Face Skin Metrics 2.0

10. Migration checklist

Attach Face Skin Analysis 3.0 to your dataset

Results only appear for applications attached to the dataset you upload to. Attach Face Skin Analysis 3.0 before you switch your upload code:

POST /api/v1/companies/{company_id}/applications/face_skin_metrics_3/attach-to-dataset/

See API for Algorithm Applications for the request details and dependency handling. Attaching Face Skin Analysis 3.0 does not remove Face Skin Metrics 2.0 from the dataset, so both can run on new batches while you migrate. Photos analysed before you attach it keep their old results only.

Face Skin Analysis 3.0 returns 15 Basic parameters for every company. The 34 Advanced parameters need the Advanced metrics add-on, which your account manager enables. See How to map old results.

How to upload

Request signed upload URL

Old method

POST /api/v1/companies/{company_id}/subjects/{subject_id}/signed_upload_url/

New method

POST /api/v3/companies/{company_id}/subjects/{subject_id}/upload/

Then as previously you need to make an actual upload request to signed upload URL.

How to send to computation

Old method

POST /api/v1/companies/{company_id}/batches/{batch_id}/compute/

New method

POST /api/v3/companies/{company_id}/batches/{image_batch_id}/compute/

How to get aux images

Old method

GET /api/v1/companies/{company_id}/images/{image_id}/aux_images/

New method

GET /api/v4/companies/{company_id}/batches/{image_batch_id}/results/

Example

{
    "batch_id": "227c206b-e0ff-425e-9f9d-9716e28ee1de",
    "subject_id": "eca81ec1-4be3-4c06-bd15-61302fa406f8",
    "dataset_id": "0c7e3d55-9b1a-4f6e-8d2a-3b5f1e9a7c42",
    "skin_consultant_app": null,
    "can_watch_predicted_images": true,
    "face_skin_analysis_3": {
        "parameters": {...},
        "all_algorithms_calculated": true,
        "predicted_images": {
            "front": {
                "original": "https://storage.googleapis.com/...",
                "anonymised": "https://storage.googleapis.com/...",
                "aligned_face": "https://storage.googleapis.com/..."
            },
            "left": {
                "original": null,
                "anonymised": null,
                "aligned_face": null
            },
            "right": {
                "original": null,
                "anonymised": null,
                "aligned_face": null
            }
        },
        "detection_errors": {
            "front": null,
            "left": null,
            "right": null
        }
    },
    "recommendation_system_for_fsa_3": null
}

To get aligned_face image (previously named as restored_selfie_rgba):

aligned_face_url = response.face_skin_analysis_3.predicted_images.front.aligned_face # equals "https://storage.googleapis.com/example-bucket/subjects/subject-id/batches/image-batch-id/images/image-id/face/aligned_face"

This is a signed URL leading to an image with aligned face based on the original uploaded selfie. If can_watch_predicted_images is false for the user who makes the request, predicted_images comes back with front, left, and right set to null.

How to get results

Old method

GET /api/v1/companies/{company_id}/images/{image_id}/metrics_and_submetrics/

New method

GET /api/v4/companies/{company_id}/batches/{image_batch_id}/results/

Results are now calculated and returned per batch, not per image. You read one response for the whole batch. Masks and predicted images inside it are still split by side: front, left, and right.

The result is now an object instead of an array. Parameters are under face_skin_analysis_3.parameters, and each key is a parameter name.

Example

{
    "batch_id": "227c206b-e0ff-425e-9f9d-9716e28ee1de",
    "subject_id": "eca81ec1-4be3-4c06-bd15-61302fa406f8",
    "dataset_id": "0c7e3d55-9b1a-4f6e-8d2a-3b5f1e9a7c42",
    "skin_consultant_app": null,
    "can_watch_predicted_images": true,
    "face_skin_analysis_3": {
        "parameters": {
            "pimples": {
                "areas": {
                    "nose": {
                        "amount": 1
                    },
                    "forehead": {
                        "amount": 3
                    },
                    "perioral": {
                        "amount": 0
                    },
                    "left_cheek": {
                        "amount": 0
                    },
                    "right_cheek": {
                        "amount": 0
                    }
                },
                "masks": {
                    "front": {
                        "original": "https://storage.googleapis.com/...",
                        "aligned_face": "https://storage.googleapis.com/...",
                        "anonymised": "https://storage.googleapis.com/..."
                    }
                },
                "amount": 4,
                "algorithm": {
                    "name": "pimples",
                    "version": 1
                }
            }
        },
        "all_algorithms_calculated": true,
        "detection_errors": {
            "left": null,
            "front": null,
            "right": null
        }
    },
    "recommendation_system_for_fsa_3": null
}

To access the count of detected pimples for a batch of uploaded images:

pimples_amount = response.face_skin_analysis_3.parameters.pimples.amount  # equals 4 pimples

How to get and show masks

Old method

GET /api/v1/companies/{company_id}/images/{image_id}/metrics_and_submetrics/

New method

GET /api/v4/companies/{company_id}/batches/{image_batch_id}/results/

Example

{
    "batch_id": "227c206b-e0ff-425e-9f9d-9716e28ee1de",
    "subject_id": "eca81ec1-4be3-4c06-bd15-61302fa406f8",
    "dataset_id": "0c7e3d55-9b1a-4f6e-8d2a-3b5f1e9a7c42",
    "skin_consultant_app": null,
    "can_watch_predicted_images": true,
    "face_skin_analysis_3": {
        "parameters": {
            "pimples": {
                "areas": {
                    "nose": {
                        "amount": 1
                    },
                    "forehead": {
                        "amount": 3
                    },
                    "perioral": {
                        "amount": 0
                    },
                    "left_cheek": {
                        "amount": 0
                    },
                    "right_cheek": {
                        "amount": 0
                    }
                },
                "masks": {
                    "front": {
                        "original": "https://storage.googleapis.com/...",
                        "aligned_face": "https://storage.googleapis.com/...",
                        "anonymised": "https://storage.googleapis.com/..."
                    }
                },
                "amount": 4,
                "algorithm": {
                    "name": "pimples",
                    "version": 1
                }
            }
        },
        "all_algorithms_calculated": true,
        "detection_errors": {
            "left": null,
            "front": null,
            "right": null
        }
    },
    "recommendation_system_for_fsa_3": null
}

To access the masks of pimples detected in the uploaded batch:

pimples_masks = response.face_skin_analysis_3.parameters.pimples.masks

The response.face_skin_analysis_3.parameters.pimples.masks object contains directional mask data for each image in the batch. It may include up to three keys: front, left, and right, depending on the views uploaded.

Each key holds URLs to pre-rendered SVG masks. Here is the example for the front mask:

front_aligned_face_pimples_mask = response.face_skin_analysis_3.parameters.pimples.masks.front.aligned_face  # a signed URL: "https://storage.googleapis.com/..."

The signed URL mask will contain an SVG element showing the mask.

Example mask:

Pimples mask svg screenshot

How does the html svg element with masks look like

When you receive a signed URL pointing to an SVG mask (e.g. aligned_face), you can overlay it directly on top of the aligned face image in your UI using standard HTML.

Example how to show SVG in your HTML using JavaScript:

<div id="svg-container"></div>

<script>
fetch("https://storage.googleapis.com/...")
  .then(response => response.text())
  .then(data => {
    document.getElementById("svg-container").innerHTML = data;
  });
</script>

How to know when results are ready

Results are calculated per batch, and Face Skin Analysis 3.0 can return them before every parameter is done. Always check face_skin_analysis_3.all_algorithms_calculated before you treat the result as complete.

Webhooks (recommended)

Configure a Webhook v3 on your dataset. The platform sends a POST to your server when a batch is processed. The webhook's version setting picks the payload shape, and new webhooks use v4 unless you set v3:

• v4: the batch is batch_id, an images object lists the uploaded photos, and the results are under face_skin_metrics_3.

• v3: the batch is image_batch_id, and the results are under face_skin_metrics_3.

In both shapes, face_skin_metrics_3 carries the same all_algorithms_calculated flag and parameters. Use the batch ID to fetch the full result with GET /api/v4/.../batches/{image_batch_id}/results/ if you need masks or predicted images.

Already use a Webhook v1? Its batch payload also carries the 3.0 results, under face_skin_metrics_3, so you can keep it. The per-photo v1 payload carries only the 2.0 metrics, so move per-photo webhooks to Webhook v3. Either way, update your handler to read the 3.0 results instead of the 2.0 metrics, and update anything downstream that reads them (CRM, analytics). See Webhooks v3 for the payload, the secret_key check, and retries.

Polling

No endpoint reports on its own whether results are ready. GET /api/v3/companies/{company_id}/batches/{image_batch_id}/ returns the batch itself, not the progress of the analysis. To know when results are complete, poll the results endpoint:

GET /api/v4/companies/{company_id}/batches/{image_batch_id}/results/

While the batch is still processing, the response can contain only the parameters finished so far, with face_skin_analysis_3.all_algorithms_calculated: false. Keep waiting; this is not an error. Stop when the flag is true, or when you give up. Most batches finish in a few seconds, so poll every 1-2 seconds for up to 15-20 seconds, with a delay between calls.

How to map old results

Face Skin Analysis 3.0 returns 49 parameters: 15 Basic parameters that are always calculated, and 34 Advanced parameters. The tables below map every one of them back to its Face Skin Metrics 2.0 counterpart, where one exists.

Face Skin Analysis 3.0 paths are relative to the face_skin_analysis_3 object returned by GET /api/v4/companies/{company_id}/batches/{image_batch_id}/results/. A path such as parameters.pores.score is read as response.face_skin_analysis_3.parameters.pores.score.

Face Skin Metrics 2.0 paths are shown in short form. selfie_v2.pores → main_metric.value means selfie_v2.pores.result.area_results[0].main_metric.value. Some 3.0 parameters map to a sub-metric of a 2.0 algorithm rather than to its main metric — those are shown as selfie_v2.<algorithm> → sub_metrics → <key>.

Basic parameters

Always returned for every batch.

Face Skin Analysis 3.0 parameter	Result path in 3.0	Face Skin Metrics 2.0 equivalent	Result path in 2.0
Perceived Age Parameter Result Scheme	parameters.age.age	Perceived Age Result Scheme	selfie_v2.age → main_metric.value
Breakouts Parameter Result Scheme	parameters.breakouts.score (also .grade)	Acne Result Scheme	selfie_v2.acne → main_metric.value
Dark Circles Parameter Result Scheme	parameters.dark_circles.score (also .grade)	Eye Area Condition Result Scheme	selfie_v2.eye_bags → sub_metrics → dark_circles_score
Eyes Age Parameter Result Scheme	parameters.eyes_age.eyes_age	Eye Age Result Scheme	selfie_v2.eyes_age → main_metric.value
Hydration Parameter Result Scheme	parameters.hydration.score (also .grade)	Hydration Result Scheme	selfie_v2.hydration → main_metric.value
Lines Parameter Result Scheme	parameters.lines.score (also .grade)	Lines Result Scheme	selfie_v2.lines → main_metric.value
Pigmentation Parameter Result Scheme	parameters.pigmentation.score (also .grade)	Pigmentation Result Scheme	selfie_v2.pigmentation → main_metric.value
Pores Parameter Result Scheme	parameters.pores.score (also .grade, .amount)	Pores Result Scheme	selfie_v2.pores → main_metric.value
Under Eye Puffiness Parameter Result Scheme	parameters.puffiness.score (also .grade)	Eye Area Condition Result Scheme	selfie_v2.eye_bags → sub_metrics → eye_bags_score
Quality Parameter Result Scheme	parameters.quality.front / .left / .right	Image Quality Result Scheme	selfie_v2.quality → main_metric.value
Redness Parameter Result Scheme	parameters.redness.score (also .grade)	Redness Result Scheme	selfie_v2.redness → main_metric.value
Sagging Parameter Result Scheme	parameters.sagging.score (also .grade)	Sagging Result Scheme	selfie_v2.lines → sub_metrics → sagging_score
Skin Type Parameter Result Scheme	parameters.skin_type.classification	Skin type Result Scheme	selfie_v2.skin_type → main_metric.value
Skin Tone Parameter Result Scheme	parameters.skintone.classification	Skin tone Result Scheme	selfie_v2.skintone-classification → main_metric.value
Uniformness Parameter Result Scheme	parameters.uniformness.score (also .grade)	Uniformness Result Scheme	selfie_v2.uniformness → main_metric.value

selfie_v2.skin_type and selfie_v2.sagging were unpublished or disabled by default in Face Skin Metrics 2.0, so many integrations never received them. In Face Skin Analysis 3.0 both are Basic parameters and are always returned.

Advanced parameters

Advanced parameters are returned only if the Face Skin Analysis Advanced metrics add-on is enabled for your company. Contact your account manager to enable it.

Face Skin Analysis 3.0 parameter	Result path in 3.0	Face Skin Metrics 2.0 equivalent	Result path in 2.0
ITA Parameter Result Scheme	parameters.ita.ita (also .classification, .l_channel, .a_channel, .b_channel)	ITA (Individual Typology Angle) Result Scheme	selfie_v2.skin_tone → sub_metrics → ita_score
Fine Lines Parameter Result Scheme	parameters.fine_lines.score (also .grade)	Wrinkles Response Scheme	selfie_v2.wrinkles → sub_metrics → fine_wrinkles_density
Deep Lines Parameter Result Scheme	parameters.deep_lines.score (also .grade)	Wrinkles Response Scheme	selfie_v2.wrinkles → sub_metrics → deep_wrinkles_density
Nasolabial Wrinkles Parameter Result Scheme	parameters.nasolabial_wrinkles.grade	—	—
Marionette Lines Parameter Result Scheme	parameters.marionette_lines.grade	—	—
Forehead Lines Parameter Result Scheme	parameters.forehead_lines.grade (also .amount, .length)	—	—
forehead_fine_lines	parameters.forehead_fine_lines.amount (also .length)	—	—
forehead_deep_lines	parameters.forehead_deep_lines.amount (also .length)	—	—
Glabellar Lines Parameter Result Scheme	parameters.glabellar_lines.grade (also .amount, .length)	—	—
glabellar_fine_lines	parameters.glabellar_fine_lines.amount (also .length)	—	—
glabellar_deep_lines	parameters.glabellar_deep_lines.amount (also .length)	—	—
Lip Lines Parameter Result Scheme	parameters.lip_lines.grade (also .amount, .length)	—	—
lip_fine_lines	parameters.lip_fine_lines.amount (also .length)	—	—
lip_deep_lines	parameters.lip_deep_lines.amount (also .length)	—	—
Under Eye Lines Parameter Result Scheme	parameters.under_eye_lines.grade (also .amount)	—	—
under_eye_fine_lines	parameters.under_eye_fine_lines.amount	—	—
under_eye_deep_lines	parameters.under_eye_deep_lines.amount	—	—
Crow's Feet Lines Parameter Result Scheme	parameters.crows_feet_lines.grade (also .amount)	—	—
crows_feet_fine_lines	parameters.crows_feet_fine_lines.amount	—	—
crows_feet_deep_lines	parameters.crows_feet_deep_lines.amount	—	—
Eye Lines Parameter Result Scheme	parameters.eyes_lines.grade (also .score)	—	—
Jowls Parameter Result Scheme	parameters.jowls.grade	Lines Result Scheme	selfie_v2.lines → sub_metrics → jowls_grade
Lacrimal Grooves Parameter Result Scheme	parameters.lacrimal_grooves.score (also .grade)	Eye Area Condition Result Scheme	selfie_v2.eye_bags → sub_metrics → lacrimal_grooves_score
Enlarged Pores Parameter Result Scheme	parameters.enlarged_pores.amount	Pores Result Scheme	selfie_v2.pores → sub_metrics → pores_number
Freckles Parameter Result Scheme	parameters.freckles.density	—	—
Moles Parameter Result Scheme	parameters.moles.amount	—	—
Melasma Parameter Result Scheme	parameters.melasma.density	—	—
Sun Spots Parameter Result Scheme	parameters.sun_spots.density (also .amount, .grade, .score)	selfie_v2.age_spots (feature-flagged in 2.0, no doc page)	selfie_v2.age_spots → main_metric.value
Pimples Parameter Result Scheme	parameters.pimples.amount (also .grade)	—	—
Papules Parameter Result Scheme	parameters.papules.amount (also .grade)	—	—
Pustules Parameter Result Scheme	parameters.pustules.amount (also .grade)	—	—
Acne Inflammation Parameter Result Scheme	parameters.inflammation.density (also .grade)	—	—
Post-acne Hyperpigmentation Parameter Result Scheme	parameters.post_acne_hyperpigmentation.density (also .grade)	—	—
Irritation Parameter Result Scheme	parameters.irritation.score (also .grade)	—	—

New & Deprecated Metrics/Parameters

New in Face Skin Analysis 3.0

Parameters with no Face Skin Metrics 2.0 counterpart:

• Lines by facial region — Forehead Lines Parameter Result Scheme, Glabellar Lines Parameter Result Scheme, Lip Lines Parameter Result Scheme, Under Eye Lines Parameter Result Scheme, Crow's Feet Lines Parameter Result Scheme, Eye Lines Parameter Result Scheme, Nasolabial Wrinkles Parameter Result Scheme, Marionette Lines Parameter Result Scheme, and their _fine_lines / _deep_lines variants

• Breakouts breakdown — Pimples Parameter Result Scheme, Papules Parameter Result Scheme, Pustules Parameter Result Scheme, Acne Inflammation Parameter Result Scheme, Post-acne Hyperpigmentation Parameter Result Scheme

• Pigmentation breakdown — Freckles Parameter Result Scheme, Moles Parameter Result Scheme, Melasma Parameter Result Scheme

• Irritation Parameter Result Scheme

• Enlarged Pores Parameter Result Scheme

Promoted to top-level parameters

These existed in Face Skin Metrics 2.0 only as sub-metrics of another algorithm. In Face Skin Analysis 3.0 they are parameters in their own right:

• Under Eye Puffiness Parameter Result Scheme — was eye_bags_score inside selfie_v2.eye_bags

• Sagging Parameter Result Scheme — was sagging_score inside selfie_v2.lines

• Lacrimal Grooves Parameter Result Scheme — was lacrimal_grooves_score inside selfie_v2.eye_bags

• Jowls Parameter Result Scheme — was jowls_grade inside selfie_v2.lines

• ITA Parameter Result Scheme — was ita_score inside selfie_v2.skin_tone

Missing in Face Skin Analysis 3.0

• Translucency Result Scheme — no direct equivalent. The closest signal is Skin Type Parameter Result Scheme.

How to turn off Face Skin Metrics 2.0

When your integration reads only Face Skin Analysis 3.0 results, detach Face Skin Metrics 2.0 from the dataset. A detached application stops running on new batches that you send with POST /api/v3/.../batches/{image_batch_id}/compute/.

Move to v3 compute before you detach. The v1 compute call, POST /api/v1/.../batches/{batch_id}/compute/, still runs detached applications. Batches you send that way keep getting Face Skin Metrics 2.0 results.

POST /api/v1/companies/{company_id}/applications/face_skin_metrics_2.0/detach-from-dataset/

curl -X POST \
    https://saas.haut.ai/api/v1/companies/{company_id}/applications/face_skin_metrics_2.0/detach-from-dataset/ \
    -H "Authorization: Bearer {token}" \
    -H "Content-Type: application/json" \
    -d '{"dataset_id": "{dataset_id}"}'

A 200 response means Face Skin Metrics 2.0 is now off for that dataset. Detach works per dataset, so repeat the call for every dataset that has Face Skin Metrics 2.0 attached. To turn it on again, attach it to the dataset again. Send the enabled_algorithms and algorithms_flags you used before: an attach without them resets the application to its default algorithms and no flags.

Before you detach:

• Detach dependent applications first. These applications need Face Skin Metrics 2.0: Recommendation Engine, Recommendation System for FSM 2.0, and the SkinGPT generated-vs-original comparison applications. The detach request does not check for them, so they would stay attached with no input. Switch Recommendation System for FSM 2.0 to Recommendation System for FSM 3. The other applications have no Face Skin Analysis 3.0 counterpart, so detach them.

• Switch your code first. Once your upload and compute calls use v3, and your results and webhook code read the 3.0 results, detach. Detaching earlier leaves your integration without results.

• Old 2.0 results are hidden from v4, not deleted. Turning Face Skin Metrics 2.0 off does not delete its results. But v4 batch and subject results, Webhook v1, and Webhook v3 with version set to v4 show results only for applications that are on. So face_metrics_2 disappears from them, for old batches too. GET /api/v1/.../images/{image_id}/metrics_and_submetrics/ still returns the 2.0 results, and attaching Face Skin Metrics 2.0 again shows them in v4 again.

• System datasets cannot be edited. The request returns a permission error for them.

• The dataset must have the application attached. Otherwise you get an application_not_attached error.

Migration checklist

• Face Skin Analysis 3.0 (face_skin_metrics_3) is attached to the dataset you upload to.

• Upload uses POST /api/v3/.../subjects/{subject_id}/upload/, and you save image_batch_id from the response.

• After the file upload, you call POST /api/v3/.../batches/{image_batch_id}/compute/.

• Results are read once per batch with GET /api/v4/.../batches/{image_batch_id}/results/, under the face_skin_analysis_3 key.

• Your code reads parameters by name and ignores names it does not recognize. See Best practices.

• You check all_algorithms_calculated before treating a result as complete, and polling has a delay between retries.

• Webhook handlers and downstream systems read the 3.0 results, not the 2.0 metrics.

• Masks and predicted images come from the results response, not from aux_images.

• Your app checks the quality parameter to ask users to retake a bad photo. See Quality Parameter Result Scheme.

• Every 2.0 parameter you use has a 3.0 counterpart in the mapping tables, and Advanced parameters are enabled if you need them.

• If you still show old 2.0 results, you read them through /api/v1/.../metrics_and_submetrics/, because v4 stops returning them after detach.

• Every batch goes through v3 compute. Then Face Skin Metrics 2.0 and the applications that depend on it are detached from every dataset you migrated.

• Nothing in your integration calls a /api/v1/ metrics or aux-images endpoint for face skin results any more, except reads of old 2.0 results.

---

Not a Haut.AI client yet?

Book a demo Face Skin Analysis 3.0 on haut.ai