It's very fun! The only change I'd make would be to use real photos of the animals instead of the AI images. It's hard to relate to them the way actual pictures would be.
Pexels is an excellent source of free, creative commons images and they have all the animals you need. Here's some geckos https://www.pexels.com/search/gecko/ :)
What was the source video for the data BTW? I'm a little unsure about some of these, like cows having 5% body fat! According to this[0] webpage a "moderate" condition cow has 19% body fat, but it has a "Body Composition Score" of 5.
The body fat content for many animal species varies a lot throughout the year based on migration, food availability, and (sometimes) hibernation. So it depends on when you measure.
Note that body fat percentage is notoriously difficult to accurately measure. The linked calculator uses the navy circumference method which gives readings ±3-4 points for most people and even further off for some. It gives me 3.5% which would be insanely low (I'm probably somewhere around 8-10 right now) because I have an unusually wide neck. To get more accurate readings you'd have to pay for a DEXA scan or a bodpod but even those can be off by a couple points.
My favourite comment on this is the following joke exchange:
Doctor: Your BMI is high I'd like you make some dietary adjustments.
Patient: BMI doesn't measure fat accurately for example professional Rugby players are marked as obese but aren't.
Doctor: Are you a Rugby player?
Patrient: No.
It's a good guideline. Don't let edge cases control your perception and let professionals do their thing with the tools they have -- they're aware of limitations that's why their professionals.
While that joke is accurate (although some football players are also obese [1] depending on their position -- Aaron Gibson was 400lbs), BMI was never meant to be applied to individuals. It was designed to be applied to populations. It does a good job of that. Also note that guidelines have historically been created against a very limited selection of race and have not appropriately adjusted for height (although BMI 2.0 corrects the height issues).
If we're just relying on the professional's opinion then they don't really need the BMI, right. They can just look at you and go "huh, you're looking a little thicc today."
BMI is a good tool for population health, a bad tool for individual health, and if it just so happens to correlate to your thiccness then you probably already know that.
I agree with parent that body composition analysis via DEXA or air displacement plethysmography is a far better metric.
I have no doubt that a carpenter can bang a nail in with a screwdriver 90% of the time, that's why they're professionals after all, but when I see it, I can't help but think "there's gotta be a better way mate."
It's really not, and it's always weird when people jump to that conclusion.
Different genetics lead to different health outcomes for the same body fat percentages - or the same BMI on a population level. Just as it's not a "social matter" that the prevalence of sickle cell is much higher in African Americans. It doesn't make sense to use health outcomes standardized against a BMI range for one population racial mix against individuals of a race not represented -- in at least two separate ways.
Hong Kong for instance sets the obesity cutoff at a BMI of 25. China and Singapore use 27.5 I believe. The WHO and various other organizations generally endorse lower cut-offs for people of Asian descent. [1]
I'm not sure if you're following what I am saying. The same body fat percentage leads to different health outcomes for members of different races in aggregate because of different genetic predispositions. Someone of one race with BMI X is likely to have a different risk profile than someone of a different race with BMI X no matter what X is, high or low. This is not a social matter. We're just talking about heath statistics, not aesthetics.
That's not to say there aren't social aspects to obesity as well (obviously) but that's not what we're talking about right now, and parent dragging that in is just a distraction.
From [1] in the comment you replied:
> Access to medications and surgery should be improved, in part by updating US indications for therapies to reflect race‐specific obesity thresholds and through inclusion of Asian American people of all subtypes with lower BMI values in clinical trials.
It is bad all around, the inaccuracies start waaaay before you get to something you could call an "edge case".
I'm 5'11, 180 lb male, that counts as overweight.
I'm 18% bodyfat. You wouldn't even note me as being particularly athletic looking if you walked by, I'm right in the middle of the bell curve for "guy who works out sometimes".
There's no way it should be flagging someone like me as overweight.
100% accurate. Caveat Emptor - the world does not understand even the smallest nuance that needs to take into consideration - in this case "guideline".
For example, the US DoD used BMI (DoDI 1308.03) to disqualify a friend of mine from joining. his actual body fat by DEXA? 7%, he was just short and very fit.
There are plenty of people who do weightlifting or have high muscle density. Probably more than 10% of people and certainly not edge cases. Doctors are basically taught like undergrads are taught, not much critical thinking involved, they just recite whatever is in the relevant document they are supposed to recite from.
I really doubt more than 10% of people do weightlifting and have high muscle density. It's likely closer to 0.1% which is why BMI is a good guideline over a group.
The problem is not so much what the doctor will tell people in person but that these same people will end up in stats and should be separated into their own category. Having a high BMI and that excludes strength trainers would make high BMI stats worse than it is already.
Those people do skew the stats by a bit, but they are outnumbered by the people who are "skinny fat", i.e. people who are very sedentary and who have high body fat but still have normal BMIs due to having very low lean body mass.
Also, when using the Navy method, be sure you read the instructions for how to measure. The “waist” measurement is unfortunately not the number on your trouser size. :)
Yeah, 3.5% would make you leaner than many Mr. Olympia competitors! And nearly at the point of negative health effects.
It's a shame when people use those electronic home scales and believe the results. Someone was so happy to tell me they were at 10% body fat, I could only smile and nod...
Most healthy people would have difficulty dropping below 10-12% without very deliberate effort. And beyond that would probably be kinda freaked out when they start to see all the veins and striations appearing.
I got down to about 6% for a bodybuilding show. It was incredible to look at, but also literally painful to bump into any hard surface, to sit on any hard surface, etc. The amount of bony protrusions you have when the body fat is gone is eye opening. I walk around at 10-13% typically, then lean down for summer and shows.
Most people at 30+ lbs further away from 'having abs' than they think. I see it time and time again.
When I dip below around 8%ish my appetite goes through the roof, all self control goes out the window to the all consuming signaling in my brain to eat eat eat
I had a DEXA scan done and for me, it showed higher body fat about five percentage points higher than my scale at home did. But I used that as a calibration point for my home scale. I've found as long as I use the home scale under similar conditions each time, mostly in the morning after voiding my bladder but before showering, that there's not much fluctuation from day to day. If I ever hit my goal on the scale, I plan to go back and get a second DEXA scan.
Two interesting things that also came from getting the scan done was that it showed that I have poor bone density so that's an issue that needed addressing (and hopefully the second scan will show improvement) and the scan was part of a package where they also did a 3d scan of my body exterior, which I was able to download and 3d print. I plan on making a half-and-half model of my body's change upon hitting my goal body fat percentage.
- my intuitive sense of animals that are thin or fat is totally off. I think of cows as being fat (ex: "fat cow") but their body fat percentage is only 6%! I think of Gorillas being super fit and muscular but they're at 31%!
- I hate that I can't hit enter after changing the body fat percentage value and have it change
- on that note I'd prefer if there wasn't a "find my match" button at all—after changing the body fat percentage value the match should immediately update
- better than that would be a sliding scale of every animal and the one that matches gets highlighted and centered
- real-world pictures would be better and even better than that would be an image of the animal and a cross section/MRI of the animal where you can see the fat distribution
Cow is not 6% body fat unless it's been starving or decided to join a body building contest. The source data is bad, probably got the data from ChatGPT.
I think it'd be cool if the input was to calculate the % (if you didn't already know) and the area below was just a timeline-like output with the different animals marked along the way. It seems most people here are more interested in exploring the output space than finding their specific animal match.
All of it? How fat is a walrus, where's _your_ source? It probably varies by subspecies, age, season, etc. Maybe they are ripped af under all that blubber?
OP's source is a YouTube video that doesn't list their source. Given that video is post ChatGPT...
This reminds me of the information stealing spam we'd see on Facebook. At least, unlike many of the questions in those, I haven't seen BMI used as a security question.
Between this today and one square minesweeper yesterday, we're batting 1.000 on funny dumb projects that make me laugh. I don't know why you made this either, but I'm glad you did, thanks!
I can understand the excitement in principle that you made an app with little effort, but there is no reason this can't be a static page with a table of animals and their average body fat percentages. Users don't need to enter any information and you don't even need a backend. They can simply go the page and find the animal from the table that most closely matches their own bodyfat percentage.
The data is also dubious. I'm not going to enter every possible number to see what comes back, but the number reported here for cows is definitely wrong. Insects have cells that can store nutrients, but not adipose tissue in the familiar mammal sense that we have. There is no way to know how much fat they're actually storing at any given moment versus other nutrients.
id rather see a table than have to enter a form. that way you don't just get "your" score, but you get to see the whole list of animals and learn about which ones have high/low body fat
The highly unscientific poll aspect is irrelevant. The question should be, who would want to know the body fat percentages of potential consumers (as identified by IP addresses, cookies, and other methods of online tracking used by the forms)? This would obviously be useful for targeted advertising, for example. And this information can be correlated and gathered in giant databases with profiles of consumers.
Data brokers do not seem to care if their data is poisoned. That's their customers' problem! Their customers don't have any way of validating it, so there's no incentive to care. Haven't you ever received spam that was based on some extremely spurious connections?
If it helps, I’m not storing any user data. There’s no backend or database, the app just looks up values in a static json. Just copied the ui code from chatgpt and put it into a react app. The only thing I’m tracking is page views!
It's very fun! The only change I'd make would be to use real photos of the animals instead of the AI images. It's hard to relate to them the way actual pictures would be.
Pexels is an excellent source of free, creative commons images and they have all the animals you need. Here's some geckos https://www.pexels.com/search/gecko/ :)
I had the same thought. AI art looks cheap, but more importantly it doesn't show if the real animal looks fat or not! Pexels looks super useful.
Can also prompt AI to create photorealistic images. Midjourney will probably let you specify famous photographer styles.
> Can also prompt AI to create photorealistic images
It can, but the ‘baby peacock’ thing has damped my enthusiasm.
https://news.ycombinator.com/item?id=41767648
What was the source video for the data BTW? I'm a little unsure about some of these, like cows having 5% body fat! According to this[0] webpage a "moderate" condition cow has 19% body fat, but it has a "Body Composition Score" of 5.
[0] https://www.grass-fed-solutions.com/body-condition-scoring.h...
cows = 5%, wasps = 15%, kangaroos = 2%
I am not an animal expert but just based on intuition this seems off...
Additionally seals have 40% body fat, but walruses have 20%?
That seems suspicious.
I didn't know my body fat so I just guestimated 20 to immediately be flashbanged by the image of a walrus.
I consider myself on the skinny side so it was a surprise to be sure.
The body fat content for many animal species varies a lot throughout the year based on migration, food availability, and (sometimes) hibernation. So it depends on when you measure.
I am seeing a lot of cool feature ideas on this thread which is amazing! I just made the repo public if anyone wants to look at the json or put in a PR with some edits :) https://github.com/amittallapragada/animal-bodyfat-calculato...
Note that body fat percentage is notoriously difficult to accurately measure. The linked calculator uses the navy circumference method which gives readings ±3-4 points for most people and even further off for some. It gives me 3.5% which would be insanely low (I'm probably somewhere around 8-10 right now) because I have an unusually wide neck. To get more accurate readings you'd have to pay for a DEXA scan or a bodpod but even those can be off by a couple points.
My favourite comment on this is the following joke exchange:
Doctor: Your BMI is high I'd like you make some dietary adjustments.
Patient: BMI doesn't measure fat accurately for example professional Rugby players are marked as obese but aren't.
Doctor: Are you a Rugby player?
Patrient: No.
It's a good guideline. Don't let edge cases control your perception and let professionals do their thing with the tools they have -- they're aware of limitations that's why their professionals.
While that joke is accurate (although some football players are also obese [1] depending on their position -- Aaron Gibson was 400lbs), BMI was never meant to be applied to individuals. It was designed to be applied to populations. It does a good job of that. Also note that guidelines have historically been created against a very limited selection of race and have not appropriately adjusted for height (although BMI 2.0 corrects the height issues).
If we're just relying on the professional's opinion then they don't really need the BMI, right. They can just look at you and go "huh, you're looking a little thicc today."
BMI is a good tool for population health, a bad tool for individual health, and if it just so happens to correlate to your thiccness then you probably already know that.
I agree with parent that body composition analysis via DEXA or air displacement plethysmography is a far better metric.
I have no doubt that a carpenter can bang a nail in with a screwdriver 90% of the time, that's why they're professionals after all, but when I see it, I can't help but think "there's gotta be a better way mate."
[1] https://www.nytimes.com/2019/01/17/sports/football/the-nfls-...
Why would race matter?
Melanin doesn't effect fat comp... does it?
Body composition varies by race.
https://pmc.ncbi.nlm.nih.gov/articles/PMC4968570/
https://www.sciencedirect.com/science/article/pii/S000291652...
They don't need to be causally connected, just be correlated.
> Why would race matter?
Because social matter
It's really not, and it's always weird when people jump to that conclusion.
Different genetics lead to different health outcomes for the same body fat percentages - or the same BMI on a population level. Just as it's not a "social matter" that the prevalence of sickle cell is much higher in African Americans. It doesn't make sense to use health outcomes standardized against a BMI range for one population racial mix against individuals of a race not represented -- in at least two separate ways.
Hong Kong for instance sets the obesity cutoff at a BMI of 25. China and Singapore use 27.5 I believe. The WHO and various other organizations generally endorse lower cut-offs for people of Asian descent. [1]
[1] https://pmc.ncbi.nlm.nih.gov/articles/PMC10108164
Sickle cell isn’t affected by social matters, unlike body fat.
I'm not sure if you're following what I am saying. The same body fat percentage leads to different health outcomes for members of different races in aggregate because of different genetic predispositions. Someone of one race with BMI X is likely to have a different risk profile than someone of a different race with BMI X no matter what X is, high or low. This is not a social matter. We're just talking about heath statistics, not aesthetics.
That's not to say there aren't social aspects to obesity as well (obviously) but that's not what we're talking about right now, and parent dragging that in is just a distraction.
From [1] in the comment you replied:
> Access to medications and surgery should be improved, in part by updating US indications for therapies to reflect race‐specific obesity thresholds and through inclusion of Asian American people of all subtypes with lower BMI values in clinical trials.
It is bad all around, the inaccuracies start waaaay before you get to something you could call an "edge case".
I'm 5'11, 180 lb male, that counts as overweight.
I'm 18% bodyfat. You wouldn't even note me as being particularly athletic looking if you walked by, I'm right in the middle of the bell curve for "guy who works out sometimes". There's no way it should be flagging someone like me as overweight.
> It's a good guideline.
100% accurate. Caveat Emptor - the world does not understand even the smallest nuance that needs to take into consideration - in this case "guideline".
For example, the US DoD used BMI (DoDI 1308.03) to disqualify a friend of mine from joining. his actual body fat by DEXA? 7%, he was just short and very fit.
GP is not questioning the usefulness of body fat percentage as a statistic (that's its own question), but the crude method of measurement.
It says mine is 1.6%, which I'd say is at least 5% out.
There are plenty of people who do weightlifting or have high muscle density. Probably more than 10% of people and certainly not edge cases. Doctors are basically taught like undergrads are taught, not much critical thinking involved, they just recite whatever is in the relevant document they are supposed to recite from.
I really doubt more than 10% of people do weightlifting and have high muscle density. It's likely closer to 0.1% which is why BMI is a good guideline over a group.
being within an "ideal" or "average" bodyfat range will easily put you as overweight on BMI.
it's a useless statistic.
Doctors are much more likely to let obesity slide than they are to nag you about BMI when you're clearly in shape in my experience.
The problem is not so much what the doctor will tell people in person but that these same people will end up in stats and should be separated into their own category. Having a high BMI and that excludes strength trainers would make high BMI stats worse than it is already.
Those people do skew the stats by a bit, but they are outnumbered by the people who are "skinny fat", i.e. people who are very sedentary and who have high body fat but still have normal BMIs due to having very low lean body mass.
I've done a couple DEXA scans recently and they have in fact confirmed that the BMI calculation for me is spot on.
I also just did a DEXA scan last week and the calculator is within ~2% so pretty good considering it's free and takes 1 minute.
Also, when using the Navy method, be sure you read the instructions for how to measure. The “waist” measurement is unfortunately not the number on your trouser size. :)
Yeah, 3.5% would make you leaner than many Mr. Olympia competitors! And nearly at the point of negative health effects.
It's a shame when people use those electronic home scales and believe the results. Someone was so happy to tell me they were at 10% body fat, I could only smile and nod...
Most healthy people would have difficulty dropping below 10-12% without very deliberate effort. And beyond that would probably be kinda freaked out when they start to see all the veins and striations appearing.
I got down to about 6% for a bodybuilding show. It was incredible to look at, but also literally painful to bump into any hard surface, to sit on any hard surface, etc. The amount of bony protrusions you have when the body fat is gone is eye opening. I walk around at 10-13% typically, then lean down for summer and shows.
Most people at 30+ lbs further away from 'having abs' than they think. I see it time and time again.
When I dip below around 8%ish my appetite goes through the roof, all self control goes out the window to the all consuming signaling in my brain to eat eat eat
Nearly? 3.5% bf is unhealthy for men and probably dead for women.
I had a DEXA scan done and for me, it showed higher body fat about five percentage points higher than my scale at home did. But I used that as a calibration point for my home scale. I've found as long as I use the home scale under similar conditions each time, mostly in the morning after voiding my bladder but before showering, that there's not much fluctuation from day to day. If I ever hit my goal on the scale, I plan to go back and get a second DEXA scan.
Two interesting things that also came from getting the scan done was that it showed that I have poor bone density so that's an issue that needed addressing (and hopefully the second scan will show improvement) and the scan was part of a package where they also did a 3d scan of my body exterior, which I was able to download and 3d print. I plan on making a half-and-half model of my body's change upon hitting my goal body fat percentage.
> DEXA
Fat calipers would be less accurate than DEXA but much more accurate than Navy circumference method, yet a lot cheaper than DEXA.
Yeah, there are a couple inexpensive midground options here. Calipers is one method, bioelectrical impedance is another.
This is neat!
I have a few thoughts:
- my intuitive sense of animals that are thin or fat is totally off. I think of cows as being fat (ex: "fat cow") but their body fat percentage is only 6%! I think of Gorillas being super fit and muscular but they're at 31%!
- I hate that I can't hit enter after changing the body fat percentage value and have it change
- on that note I'd prefer if there wasn't a "find my match" button at all—after changing the body fat percentage value the match should immediately update
- better than that would be a sliding scale of every animal and the one that matches gets highlighted and centered
- real-world pictures would be better and even better than that would be an image of the animal and a cross section/MRI of the animal where you can see the fat distribution
Cow is not 6% body fat unless it's been starving or decided to join a body building contest. The source data is bad, probably got the data from ChatGPT.
I think it'd be cool if the input was to calculate the % (if you didn't already know) and the area below was just a timeline-like output with the different animals marked along the way. It seems most people here are more interested in exploring the output space than finding their specific animal match.
Source data is very wrong. A walrus is not 20% body fat.
All of it? How fat is a walrus, where's _your_ source? It probably varies by subspecies, age, season, etc. Maybe they are ripped af under all that blubber?
OP's source is a YouTube video that doesn't list their source. Given that video is post ChatGPT...
I am not 10% BF, but was very surprised to see the African Elephant come up for that!
I'm skeptical of the source data.
Men and women will have very different body fat percentage level for similar look. In general it's something like women's = men's +8-9%?
A man is considered lean around 15%, a woman is lean around 20%.
Men: https://images.app.goo.gl/Gn5pBBN2FB22YGsz7
Women: https://images.app.goo.gl/wAP5VDfCzKiWWYwG9
I clicked through all the ones from 1% (crab) to 25% (sheep). I honestly can't see any pattern here
"Your closest match is a lion".... alright, I'll continue drinking coffee and staring at the screen all day then.
This reminds me of the information stealing spam we'd see on Facebook. At least, unlike many of the questions in those, I haven't seen BMI used as a security question.
Between this today and one square minesweeper yesterday, we're batting 1.000 on funny dumb projects that make me laugh. I don't know why you made this either, but I'm glad you did, thanks!
Here's another that I discovered today: not a hobby project but a personal webpage that got me rotfl. https://cookie.engineer/
It needs to push state the bmi to the url so that I can send it to people.
It could also be fun if one could fill out the age, weight, height and gender to calculate the bmi.
Now that I think about it it would be hilarious to use sliders and have a cartoon person populated with the data generated in real time.
Whoa, I was shocked that an elephant was sub 10% body fat...now to call some one an elephant and explain it's a compliment!
What's the source on the images? are they AI-generated?
I was wondering the same thing, and checking the images they are hosted on https://craiyon.com/
Craiyon - Create AI Art with our free AI image generator.
There's a crayon in the bottom right corner of each of the images. Perhaps someone recognizes it?
The crayon is indeed a big hint: https://www.craiyon.com/
Yea they're all AI generated! I used this site https://www.craiyon.com/ They have a free tool for image generation
They look AI generated to me.
I can understand the excitement in principle that you made an app with little effort, but there is no reason this can't be a static page with a table of animals and their average body fat percentages. Users don't need to enter any information and you don't even need a backend. They can simply go the page and find the animal from the table that most closely matches their own bodyfat percentage.
The data is also dubious. I'm not going to enter every possible number to see what comes back, but the number reported here for cows is definitely wrong. Insects have cells that can store nutrients, but not adipose tissue in the familiar mammal sense that we have. There is no way to know how much fat they're actually storing at any given moment versus other nutrients.
Users could still enter the information. But instead of input form it would be browsers 'find in page' search bar.
I consider myself slim but so I was pretty amused to see my spirit animal is the pig. Great idea.
id rather see a table than have to enter a form. that way you don't just get "your" score, but you get to see the whole list of animals and learn about which ones have high/low body fat
I was completely confused to see that a Walrus has only 2% points more body fat than a Gecko.
Geckos store fat in their tails
https://th.bing.com/th/id/OIP.UYKInXyy0_Vhs1qqTziRGwAAAA?rs=...
That's one big fat tail, if I've ever seen one.
How long did it take you to make?
The cutworm moth is 72% body fat. Crazy!!!!
Héhéhé. That is fun!
Having insects in there feels a bit odd to me. Not exactly relatable.
Err hornet.
This is fun! Take my internet points!
Good job.
Cat master race reporting in.
Great site, thanks for sharing!
I always suspect that viral online forms like this are simply fronts for collecting personal information to sell.
Who would want to buy the results of a highly unscientific poll about body fat percentages?
The highly unscientific poll aspect is irrelevant. The question should be, who would want to know the body fat percentages of potential consumers (as identified by IP addresses, cookies, and other methods of online tracking used by the forms)? This would obviously be useful for targeted advertising, for example. And this information can be correlated and gathered in giant databases with profiles of consumers.
Well, maybe it's just me, but I can see their data being posioned by checking different numbers to see all animals :)
Data brokers do not seem to care if their data is poisoned. That's their customers' problem! Their customers don't have any way of validating it, so there's no incentive to care. Haven't you ever received spam that was based on some extremely spurious connections?
I did that too but the first number I inputted was my percentage. I'd say most people would do the same thing.
People trying to sell people ozempic and other glp-1 compounds.
If it helps, I’m not storing any user data. There’s no backend or database, the app just looks up values in a static json. Just copied the ui code from chatgpt and put it into a react app. The only thing I’m tracking is page views!
Nothing against using what you want, but isn't React bit overkill here? (For one input and one image).
Now they know I'm 100% body fat
The design encourages trying multiple values, and most people visiting the site probably don't even know their bf%