Lean vs Foodvisor. The photo scan pioneer versus the only app that rebuilds your TDEE continuously.

Foodvisor sees your plate, not your real expenditure

If you are reading this, you have probably already installed Foodvisor. You chose it precisely for what nobody did before it: photographing your plate and letting the AI recognize the chicken, the rice, the sauce, and estimate the portions without pulling out the scale. You entered your weight, height, age and sex, and picked your activity level from a static list. The app showed you a calorie target, say 2,250 kcal to lose weight.

You played along. You scanned your meals, corrected the portions when the AI hesitated, kept a clean log day after day. For the first 6 weeks, it works. You lose. You are happy. Then around week 8, the scale freezes. You tighten the screw. You drop to 2,000 kcal. Again, nothing moves.

Imagine Foodvisor shows you a TDEE of 2,500 kcal. You eat 2,250 (a theoretical deficit of 250 kcal). But in reality, your TDEE has dropped to 2,200 kcal because of metabolic adaptation. You are in a 50 kcal surplus without knowing it. No chance of continuing to lose, even with the cleanest input on the market.

The Foodvisor promise is clear and kept: you know what is on your plate without weighing anything. That is precious. What Foodvisor does not do is recompute your expenditure as the weeks of deficit go by. And that is exactly where the promise stops, even though it is the lever that makes you lose weight.

The 1990 BMR formula, without measured body fat

To calculate your basal metabolic rate (BMR, the energy you burn at rest), Foodvisor starts from your profile: weight, height, age, sex. That is the logic of almost every mainstream calorie tracker, inherited from population formulas like Mifflin-St Jeor. And let's be honest: it is better than the Harris-Benedict 1919 that other apps still use.

Mifflin-St Jeor dates from 1990 (PubMed 2305711). The sample is large (498 subjects), the indirect calorimetry methodology is serious, the formula is calibrated on a modern population: 10 × weight (kg) + 6.25 × height (cm) − 5 × age − 161 (women) or +5 (men).

The problem is not the chosen formula. The problem is what no formula of this type can see: it only takes your weight into account. Not your body fat. Not your lean mass. No field in the Foodvisor onboarding asks for your body fat percentage, and no measurement exists in the app.

Yet since the 1980s, we've known that fat mass burns very little energy compared to the rest of the body. The liver, brain, heart, kidneys, and especially muscles are the real energy sinks. Fat mass is inert. Someone at 30% bodyfat does not burn anywhere near as much as someone at 10% bodyfat, even at identical weight.

Frankenfield 2013 (PubMed 23631843) compared Mifflin-St Jeor to reference indirect calorimetry across obese and non-obese cohorts. Result: 87 % accuracy in non-obese subjects, and only 75 % in obese subjects. A more recent study (PMC11820646) shows that for BMIs above 35, Mifflin is off by 250 to 315 kcal per day. That’s a full snack’s worth of error in your deficit calculation.

500 kcal is not nothing. If the app tells you "your BMR is 2,500" when it is actually 2,000, everything downstream is wrong: your deficit target, your projected weekly loss, your macro split calculated as a percentage of TDEE. And no plate photo, however well recognized, corrects a wrong target.

Real body fatPhoto, 5 seconds

Partial conclusion: if an app calculates your BMR only from your weight, height, age and sex, the result cannot be individualized. It is mathematically impossible. Even with the best plate recognition as input.

The activity multiplier, chosen once and for all

This is where it gets serious. And it’s probably the point nobody ever explained to you.

Once your BMR is estimated, Foodvisor has to get to your total TDEE. The TDEE is BMR plus everything else : expenditure from steps, daily activities, sport, and digestion. Everything that isn't basal metabolism.

How does Foodvisor do that? Like almost every tracker: it asks you, at sign-up, to pick your activity level from a static list. In sports science these factors are called PAL levels (Physical Activity Level), it’s just a multiplier applied to your BMR:

• Sedentary (PAL 1.2): desk job, little walking
• Lightly active (PAL 1.375): occasional walking
• Active (PAL 1.55): sports 3 to 5 times per week
• Very active (PAL 1.725): intense sport almost daily
• Extremely active (PAL 1.9): very intense sport or a physical job

And depending on your choice, the app multiplies your BMR by the matching coefficient. That's it. That's all there is behind your daily calorie target. A box YOU ticked only once at signup. Often six months ago. Untouched since.

And here’s the silent trap: this approximation is wildly imperfect. The difference between a day stuck on the couch watching Netflix and a day at Disneyland walking 15 km with your kids over 1,000 kcal. None of the 5 boxes captures that.

Foodvisor does know how to track your activity: the app can count your steps and log your workouts. But that data mostly serves to display your activity, not to rebuild a full TDEE: your calorie target stays sitting on the multiplier chosen at onboarding, and sports expenditure gets added on top without NEAT and EAT being cleanly separated.

You can’t reduce your activity level to a static box. You might be active in weeks when you barely work from home, and sedentary in weeks when you never leave the office. You might be active in summer and sedentary in winter. You might be active from Tuesday to Friday and sedentary on weekends.

Which box will you tick this week? The truth is that none of the 5 will be correct. So Foodvisor will give you a TDEE that is systematically decorrelated from reality.

The key point of this article: even with a perfect BMR formula, the static multiplier alone would break everything. You cannot estimate a NEAT, EAT and TEF with a single multiplier on top of BMR. Conceptually absurd.

You get the idea: a BMR formula without body fat, plus a static approximation of everything else, leaves you very little chance of reaching your goals over 3 to 6 months. However clean the input on the plate side.

See your real TDEE, broken down into BMR + NEAT + EAT + TEF. Free download.

Metabolic adaptation, never modeled

This is the final boss. The most subtle concept. And probably the most important.

When you are in a calorie deficit, your body understands it is receiving less energy than before. To protect itself, it switches into economy mode. Exactly like your iPhone's low power mode: everything keeps working, but using less energy. Your BMR drops. Your NEAT drops. Your EAT drops.

This is called metabolic adaptation. The scientific literature is clear and reproducible: Müller 2015 (PubMed 26399868, Minnesota revisit), Doucet 2001 (PubMed 11430776), Nunes 2020 (PMC7484122) over 6 weeks of deficit. Here are the numbers:

• Deficit of −250 kcal per day, over 2 to 8 weeks: adaptation of 5 to 10% (TDEE drops to 90-95 % of the initial level)
• Deficit of −500 kcal per day: 10 to 15% adaptation (TDEE drops to 85-90 %)
• Deficit of −750 kcal per day: 15 to 25% adaptation (TDEE drops to 75-85 %)

Lean convention: 100 % = optimal, 90 % = 10 % adaptation. And since NEAT, EAT and TEF all depend directly on the BMR, almost the entire TDEE is impacted.

Concretely: if you planned a 10 % deficit on a TDEE of 2,500 (eating 2,250 per day), and your body adapts by 10 %, your real TDEE has dropped to 2,250. You’re at maintenance. You stop losing.

The trap is how insidious it is. At first, you lose weight. You’re happy. You keep going. But week after week, the adaptation stacks. And at some point, without changing anything in your tracking, you stop losing.

95% of people go through this without understanding. They blame their willpower. They blame their "broken metabolism". They jump into harsher diets, which makes adaptation worse. Spiral.

Foodvisor never calculates metabolic adaptation. It gives you a fixed calorie target as long as you do not manually update your weight and activity level. You can scan your plates with exemplary consistency, but when you plateau after 6 weeks of cutting, the app has no idea why.

How Lean fixes each of the 3 problems

Foodvisor a posé un standard sur le scan photo d’un plat, et sa reconnaissance de la cuisine française reste excellente. Le problème n’est pas ce qu’il voit dans ton assiette, c’est ce qu’il ne voit pas de ton corps : la dépense reste estimée par une formule de population multipliée par un niveau d’activité. Lean fait les deux : scan photo IA and mesure de chaque composant du TDEE (BMR + NEAT + EAT + TEF) plus l’adaptation métabolique. Voici le détail.

Step 1AI BodyScan

Step 2BMR recalculated

BMR on real bodyfat

Proprietary patented model, built on lean mass

Compter parfaitement les calories entrantes ne sert à rien si les calories sortantes sont fausses de 300 kcal. Lean calcule le métabolisme sur ta lean mass, la seule qui consomme réellement au repos, et non sur ton poids brut.

The AI BodyScan applique à ton corps ce que Foodvisor applique à ton assiette : une photo, un modèle entraîné sur une banque de scans DEXA, ton bodyfat en quelques secondes, à refaire chaque semaine.

Pas de pince à pli cutané, pas de balance à impédance, pas de DEXA. La même simplicité qu’un scan de repas, appliquée à ta composition corporelle.

No activity coefficient

NEAT, EAT, TEF calculated separately

NEAT. Tes pas réels arrivent via HealthKit (iOS) ou Google Fit (Android) et deviennent des calories selon ton métabolisme. C’est le poste le plus variable de la journée, et celui qu’un niveau d’activité déclaré aplatit complètement.

EAT. Chaque séance est chiffrée par MET sur ton temps d’effort réel, temps de repos exclus. Compter une heure de musculation comme une heure de course fausse le bilan de plusieurs centaines de kcal par semaine.

TEF. Foodvisor identifie tes aliments, Lean en tire le coût de digestion : 20 à 30 % des calories pour les protéines, 5 à 10 % pour les glucides, 1 à 3 % pour les lipides, au lieu du forfait de 10 % appliqué partout.

NEAT

EAT

TEF

Automatic metabolic adaptation

A world first on a consumer app

L’adaptation métabolique. Aucun scan de repas ne détecte que ton métabolisme a ralenti de 12 % après huit semaines. Lean l’estime selon les fourchettes publiées (Müller 2015, Doucet 2001) et corrige ton objectif en conséquence.

Au-delà de 10 à 15 % d’adaptation, l’app peut conseiller un retour à la maintenance pour relancer le métabolisme avant de repartir.

Aucun multiplicateur d’activité à choisir. Chaque composant est mesuré, semaine après semaine.

MethodMetabolic adaptation

Lean versus Foodvisor, criterion by criterion

An honest read of each app's strengths and weaknesses. No criterion touches price.

Lean

Foodvisor

Criterion

Lean

Foodvisor

BMR formula

Proprietary patented model (lean mass)

Population formula (Mifflin-St Jeor 1990)

Uses bodyfat

Yes, measured in the app

No, weight-height-age-sex only

Bodyfat measured inside the app

BodyScan AI via photo

No

NEAT (steps, non-exercise activity)

Computed from real steps every day

− Steps tracked, but the target sits on the static multiplier

EAT (exercise expenditure)

Per sport via MET, effective time

− Workouts logged, added to a frozen target

TEF (digestion)

Calculated from macros, integrated into the TDEE

Not integrated into the expenditure calculation

Metabolic adaptation

Automatic, week by week

No

Activity multiplier to pick

No, computed on real data

Yes, static multiplier chosen at sign-up

AI photo scan of a meal

Yes, unlimited

Yes, the historical pioneer (2018)

Visual food recognition

Yes, modern AI photo scan

The French market reference, years of training

Barcode scan

Yes

Yes

Food database

USDA + OpenFoodFacts, curated

Large database, French products well covered

Coaching by dietitians

Out of scope, the app guides you via the Progression Pyramid

Yes, registered dietitians (dedicated offer)

Calorie deficit recommendation

Adapted to real TDEE

Fixed target, manual recompute required

French app

Yes

Yes, born in Paris

Reputation and audience size

4.7/5, 10,000+ users, young FR app

Recognized photo scan pioneer, strong brand awareness in France

Business model

Premium, 7-day free trial on the annual subscription

Free version, coaching as an option

3 ways to track a meal

Foodvisor a prouvé qu’une photo pouvait remplacer une saisie manuelle. Lean reprend cette idée et l’élargit : trois méthodes d’enregistrement selon le contexte, pour tenir sur la durée.

Method 1Food database

Method 2Barcode

Method 3AI photo scan

1. Database search. Curated base, USDA + OpenFoodFacts. No community noise, no "Roast chicken" entered 47 times by 47 different users with 47 different values.
2. Barcode scan. Standard. You scan your pasta box, you get the macros.
3. AI photo scan of a meal. You photograph your plate, the AI detects the foods, you get calories and macros per food. The reflex you already have if you come from Foodvisor: you keep it as is.

Le scan photo IA de Lean joue le même rôle que celui de Foodvisor pour les repas pris dehors. La différence se situe ailleurs : ce que Lean fait ensuite de ces calories, en les confrontant à une dépense mesurée et non estimée.

Au-delà du repas, Lean affiche un TDEE qui se met à jour pendant la journée selon tes pas. Scanner parfaitement une assiette face à un objectif calorique figé ne suffit pas.

Et au-dessus, la Pyramide de Progression :

What Foodvisor does better

Lean is not perfect, and Foodvisor has several real strengths that deserve recognition. An honest read, criterion by criterion, on the axes where the pioneer stays ahead. None of these axes is secondary: they are real pillars of the Foodvisor promise.

Axis

Foodvisor

Lean

Visual recognition of a plate

9,5

8,0

Automatic portion estimation

9,0

7,5

Human coaching (registered dietitians)

9,2

2,5

Photo scan first-mover (French market)

9,5

4,0

Honest read. On plate recognition, Foodvisor created the category in France in 2018 and its AI has years of training ahead: food identification, portion estimation without a scale, handling of mixed dishes. It is its historical playground and it remains the reference there. On human support, Foodvisor offers follow-up by registered dietitians directly in the app: Lean does not offer it, and does not claim to replace it. Lean's AI photo scan is modern, unlimited and largely sufficient for daily use, but Lean does not claim first-mover status on that ground.

If your main angle is the most seasoned plate identification possible, or human coaching built into the app, Foodvisor is more relevant than Lean. If your angle is the precision of the TDEE calculation, body fat measured every week via BodyScan AI, and automatic metabolic adaptation, that is exactly what was just demonstrated in the previous 3 sections. Some people run both apps in parallel while they decide, and that is entirely defensible.

Who Lean is built for

4 profiles. If you recognise yourself in at least one, Lean is probably built for you.

Foodvisor remains more relevant for : human coaching by registered dietitians directly in the app, and the most seasoned plate recognition on the French market. Expenditure calculation precision and metabolic adaptation are simply not part of its core promise.

Switching from Foodvisor to Lean (or using both) in 3 minutes

Important note. Lean does not automatically import your Foodvisor history, nor your favorite foods. If your follow-up with a Foodvisor dietitian matters to you, nothing stops you from keeping both during the transition: Foodvisor for the human support, Lean for TDEE and daily tracking. HealthKit / Google Health Connect sync, on the other hand, takes over immediately for your steps and activity history.

Download Lean and start the BodyScan AI right now. Free sign-up.

What Lean does that Foodvisor does not (on expenditure)

Six features centered on expenditure, nowhere to be found in Foodvisor. They all stem from the same principle: calculate every TDEE component precisely, do not approximate it.

You install the app for free, you try it without commitment, then you decide if the tool fits your goal.

Frequently asked questions

The plate is solved. Expenditure is not.

This is not Foodvisor versus Lean as a marketing match. It is input versus output, two halves of the same equation.

Foodvisor solved the left half: knowing what you eat, without a scale, thanks to the most seasoned photo scan on the French market. But for the right half, your expenditure, Foodvisor relies on a 1990 population formula without measured body fat, a frozen activity multiplier you tick once at sign-up, and no metabolic adaptation. The combination of the three makes any precise calorie tracking impossible beyond a few weeks of cutting. It is mathematical.

Lean was built for that half: BMR based on the real bodyfat (measured by BodyScan AI) via a proprietary patented model, NEAT from real steps, EAT per sport via MET, TEF from macros, plus metabolic adaptation that modulates BMR week after week. Every component calculated precisely, with no magic coefficient. And the photo reflex you picked up with Foodvisor, you keep it: the AI photo scan is built in, unlimited.

If you tried Foodvisor seriously and did not get the results you hoped for on your cut, the problem is not you, nor the photo. The problem is the frozen TDEE under the hood. Change the engine, keep the reflex.