How to Log Cooking Oil, Butter and Sauces in a Photo Calorie App (Hidden Calories Guide)

Callie is our app. This guide shows how to log hidden fat in any photo tracker, and we name the other apps that handle it well. Calorie values come from USDA FoodData Central, and the full source list is at the bottom.

The 30-second answer

Photo calorie apps undercount meals mainly because they can't see fat. Oil soaked into a stir-fry, butter melted into rice and dressing tossed through a salad leave almost no trace in a photo, and at 9 calories per gram they add up fast. The fix is to add the fat yourself before the log saves: about 1 tablespoon of oil (119 calories, USDA) for a home stir-fry, 2 teaspoons for a pan-fried portion, 2 tablespoons of dressing for a salad, and an extra tablespoon when a restaurant cooked it. Use an app with a review step, like Callie, Cronometer Gold or MacroFactor, so the correction goes in before the meal hits your total.

Why cameras miss fat

A photo records shape, colour, texture and rough area. That's enough to tell chicken from tofu and to guess how much rice is on the plate. It doesn't tell anyone how the food was cooked.

Here is what stays invisible:

• Absorbed oil. A stir-fry cooked in two tablespoons of oil looks almost the same as one cooked in a teaspoon. The difference is about 200 calories.
• Melted butter. Butter in mashed potatoes, on toast after it soaks in, or stirred into rice or pasta.
• Dressings and sauces already mixed in. A Caesar salad tossed in the kitchen shows leaves and croutons. The dressing is coating them.
• Fat in the cooking base. Ghee or oil in a curry, the oil a tadka is bloomed in, cream in a soup.

Fat has 9 calories per gram, against 4 for protein and carbohydrate. It's the densest macro and the hardest one to see, which is a bad pairing for a camera.

The research points the same way. In a 2026 NIH abstract we covered here, researchers logged 102 meals of known composition in four photo apps. Every app came in low, by roughly 250 to 345 calories per meal on average, and fat was off by about 30 grams. Thirty grams of fat is around 270 calories, so most of the gap was fat. The meals were logged from the photo alone with no corrections, which is the worst case. It is still the case you get if you never open the review screen.

If your logs look right and your weight won't move, missing fat is one of the first things to check. We go through the rest of the list in calorie deficit but not losing weight.

Hidden-fat calorie reference table

These are the numbers to keep in your head, or in your app's saved foods. Per-100 g values are from USDA FoodData Central (SR Legacy). Where USDA lists a tablespoon weight we used it. Where it doesn't, we multiplied the per-100 g value by a typical tablespoon weight, noted in the table.

Feature	Per 1 tbsp	Per 1 tsp	Per 100 g	Tbsp weight used
Olive oil	119 kcal	40 kcal	884 kcal	13.5 g
Canola (vegetable) oil	124 kcal	41 kcal	884 kcal	14 g
Sesame oil	120 kcal	40 kcal	884 kcal	13.6 g
Butter, salted	102 kcal	34 kcal	717 kcal	14.2 g
Ghee (butter oil, anhydrous)	112 kcal	37 kcal	876 kcal	12.8 g
Mayonnaise, regular	94 kcal	31 kcal	680 kcal	13.8 g
Caesar dressing, regular	80 kcal	27 kcal	542 kcal	14.7 g
Ranch dressing, regular	64 kcal	21 kcal	430 kcal	15 g
Pesto, ready-to-serve	About 68 kcal	About 23 kcal	426 kcal	16 g (our estimate)
Peanut sauce (peanut butter, water, soy sauce)	46 kcal	15 kcal	257 kcal	18 g

Two things stand out. First, every pure oil is the same 884 calories per 100 g. Switching from canola to olive oil changes the flavour and the fat profile. It does nothing for your calorie count. Second, creamy dressings are lighter per spoon than oil because they contain water, but people pour more of them. A restaurant ramekin often holds several tablespoons.

Default add-ons by cooking method

When you didn't measure, a sensible default beats logging zero. These are the amounts we add in our own logs. They're our working estimates, not published standards, so swap in your own once you've measured your usual pour a few times.

Deep-fried food is the exception. How much oil a piece of food absorbs depends on the batter, the temperature and how long it sat in the fryer, and there's no good single number. Search for the fried version of the food, like "fried chicken thigh, battered", and log that instead of a plain item plus oil. Those entries describe the food as served, with the absorbed oil already counted, which gets you closer than any rule of thumb.

For regional home cooking, where the fat is part of the dish, see our sibling guides on tracking Indian food and logging homemade meals and recipes.

How to measure oil at home

You don't need to weigh everything. Oil is the one thing worth measuring because the difference between a splash and a glug is a few hundred calories. Two methods work.

Measure before you pour. Pour the oil into a measuring spoon, then into the pan. It takes three seconds. Doing it for a week teaches you what your usual pour actually is, and many people find it's two or three times what they guessed.

Weigh the bottle. Put the oil bottle on a kitchen scale before you cook and again after. The difference in grams is what went into the pan. Multiply by 9 for calories, so 20 g gone is about 180 calories. If the dish serves four, divide by four. This also works for butter, if you weigh the block.

If you batch-cook, log the whole pot as a recipe once and save it. Then each portion carries its share of the oil without you thinking about it. Our guide on homemade meals and recipes walks through that.

On cooking spray: labels use a 1/3-second spray of about 0.25 g as the serving. That's under 5 calories, and FDA labelling rules allow a serving under 5 calories to be declared as zero. A real three-second spray is around 2 to 3 g of oil, roughly 20 calories by our arithmetic. Not much, but not zero.

Do you need a food scale for the rest of the meal? Mostly no, if your app gets portions close. We cover where a scale pays off in do you need a food scale with an AI tracker.

How to add hidden fat in Callie

Callie never logs a photo scan automatically. The scan opens a review screen first, and that's where the fat goes in.

1. Take the photo. Use the original camera photo, not a screenshot, so Callie can read the depth and camera metadata it uses for portion size.
2. Check the review screen. You'll see each detected food with its portion, calories and macros.
3. Add the fat. Tap to add an item and type or say "1 tbsp olive oil" or "2 tsp butter". If Callie already found the butter on your toast but the amount looks light, edit the portion instead.
4. Confirm. Nothing reaches your daily total until you do.

You can also skip the photo and describe the meal by voice or text, fat included: "chicken stir-fry with rice, cooked in a tablespoon of sesame oil." For a dish you cook every week, save it as a meal or recipe once with the oil already in it, and re-log it in one tap. Callie handles home cooking from many cuisines, including Indian, Arab and Asian dishes, from a photo or a description.

Callie doesn't prompt you about hidden fat. You add it yourself. If you want the app to remind you, Cronometer Gold does that.

Pricing is 3 AI logs to try it, then $34.99 a year in the US. Details are on our pricing page.

How other apps handle hidden fat

Cronometer Gold has the most direct answer to this problem. Its photo logging matches foods to Cronometer's database, then shows Ingredient Suggestions, which Cronometer says are there to catch commonly missed extras such as butter on toast or cooking oil. You review, adjust and add before anything is logged. Cronometer itself notes that photos can miss sauce under a bun, milk in coffee and oil used in cooking. Gold is $59.99 a year. We compare it with Callie in Callie vs Cronometer.

MacroFactor breaks an AI photo or description into separate ingredients and puts them in an editable Plate view. Its help docs say nothing is saved until you tap Log Foods, so you can add "1 tbsp olive oil" to the plate first. It costs $71.99 a year. See Callie vs MacroFactor.

Cal AI adds a scanned meal to your day straight away in our testing. You can still edit the entry, but only after it has counted. That's workable if you remember every time. We look at its accuracy in more detail in is Cal AI accurate.

SnapCalorie takes voice notes such as "two eggs, tablespoon of butter", which is a good way to get fat in. Its listing says you adjust predictions after logging.

Feature	Callie	Cronometer Gold	MacroFactor	Cal AI	SnapCalorie
Review screen before a photo is logged					Edit after
Add oil or butter before the log saves				After logging	After logging
Prompts for commonly missed extras	You add them	Ingredient Suggestions	Not documented	Not documented	Not documented
Describe the meal by voice or text, fat included		Voice	Describe	See app	Voice notes
Save a home recipe with the oil in it				See app	See app
Annual US price	$34.99	$59.99	$71.99	Up to $29.99 listed	$149.00

Restaurant food: assume more fat

Restaurant kitchens cook for flavour, and fat carries flavour. Researchers have measured this with bomb calorimetry, which burns a meal to measure its energy directly.

• A 2018 BMJ study measured 223 frequently ordered meals in five countries and compared them with US data. The weighted mean for a US restaurant meal was 1,088 calories.
• A 2011 JAMA study tested 269 restaurant items against their stated calories. Overall the stated values held up, but 19% of items had at least 100 calories more than stated, and the gaps were largest for sit-down restaurant dishes listed under 600 calories.

Neither study isolates cooking fat, so there's no clean "restaurants use X% more oil" figure we'd stand behind. Our working rule is an estimate: log the restaurant dish as you would a home version, then add one more tablespoon of oil or butter for anything sautéed, fried, glazed or in a cream or butter sauce. For salads, assume the dressing was generous, or ask for it on the side and log what you use.

If a chain publishes nutrition information, use their number. It already includes the fat, and the JAMA data suggests chain figures are close on average.

When another app is the better choice

• You want the app to remind you about fat. Cronometer Gold's Ingredient Suggestions flag missed butter and oil, and its database has lab-analysed data for 80-plus nutrients. It costs more than Callie, at $59.99 a year.
• You cook from detailed recipes and train seriously. MacroFactor's ingredient-level Plate view and its adaptive targets are excellent, at $71.99 a year.
• You mostly eat at chain restaurants. MyFitnessPal's database covers 380-plus restaurant chains, and chain entries already include the fat. Its photo scan is Premium-only.

Callie fits best if you want photo, voice and text logging with a review step in one app, at $34.99 a year, and you're willing to add the fat yourself.

Related reading

• AI calorie app accuracy study 2026, on the NIH abstract and why fat dominates the error.
• How to track calories accurately, on what to weigh and what to estimate.
• How to track calories from a photo, on taking photos that give any app a fair chance.
• Is Cal AI accurate?, on where Cal AI's estimates drift.
• Do you need a food scale with an AI tracker?, on where a scale earns its place.

Sources

1. USDA FoodData Central, SR Legacy. "Oil, olive, salad or cooking"; "Oil, canola" (FDC 172336); "Oil, sesame, salad or cooking"; "Butter, salted"; "Butter oil, anhydrous" (FDC 173412). Accessed October 12, 2026. https://fdc.nal.usda.gov/
2. USDA FoodData Central, SR Legacy. "Salad dressing, mayonnaise, regular" (FDC 171009); "Salad dressing, caesar dressing, regular" (FDC 169055); "Salad dressing, ranch dressing, regular" (FDC 173592). Accessed October 12, 2026. https://fdc.nal.usda.gov/
3. USDA FoodData Central, SR Legacy. "Sauce, pesto, ready-to-serve, shelf stable" (FDC 171580); "Sauce, peanut, made from peanut butter, water, soy sauce" (FDC 174070). Accessed October 12, 2026. https://fdc.nal.usda.gov/
4. Charles O, Hengist A et al. (2026). "Accuracy of photo-based dietary tracking applications." NUTRITION 2026, American Society for Nutrition. Press release: https://www.eurekalert.org/news-releases/1136415
5. Roberts SB, Das SK, Suen VMM et al. (2018). "Measured energy content of frequently purchased restaurant meals: multi-country cross sectional study." BMJ 363:k4864. https://pmc.ncbi.nlm.nih.gov/articles/PMC6290458/
6. Urban LE, McCrory MA, Dallal GE et al. (2011). "Accuracy of stated energy contents of restaurant foods." JAMA 306(3):287–293. https://pubmed.ncbi.nlm.nih.gov/21771989/
7. Code of Federal Regulations, 21 CFR 101.9, "Nutrition labeling of food." https://www.ecfr.gov/current/title-21/chapter-I/subchapter-B/part-101/subpart-A/section-101.9
8. Cronometer. "Photo Logging a meal" (accessed October 12, 2026). https://cronometer.com/blog/photo-logging/
9. Cronometer Support. "Mobile Photo Logging" (accessed October 12, 2026). https://support.cronometer.com/hc/en-us/articles/39013533811092-Mobile-Photo-Logging
10. MacroFactor Help Center. "AI Food Logging." https://help.macrofactorapp.com/en/articles/258-ai-food-logging
11. SnapCalorie FAQ (accessed September 28, 2026). https://www.snapcalorie.com/faq.html
12. Cal AI App Store listing (accessed September 28, 2026). https://apps.apple.com/us/app/cal-ai-calorie-tracker/id6480417616