Question for tech person with good language skills
3I’m having trouble interpreting this statement as to actual numbers.
From a Google developer page:
lossy WebP also supports transparency, typically providing 3× smaller file sizes compared to PNG
So if the “comparable PNG” file size is 120 kB, what would the “lossy WebP” size be?
Would it follow then that the PNG file is 3x larger? Or what?
Anybody know examples with actual numbers comparing the sizes?
I’m not familiar with using any image-editing applications, but might need to become such soon, so I don’t have background for this question.
- 5 comments, 12 replies
- Comment
I hate it when people use the “times” when explaining smaller or fewer. If I have a quantity of 1,000 of something and then say I now have 3 times fewer of that item, how many is that? The math would say you need to multiply by 3 so that gives you 3,000 and now it says fewer so you subtract. That means you’re left with -2,000, right?
They should state 1/3rd less or 33% less or 333 fewer or just say you’re left with only 667 now.
When someone says 3X smaller or fewer I would assume to reduce by 1/3rd but who knows because in my example, you cannot multiply 667 by 3 to get the original 1,000 so you’d have to reduce it by 2/3rds to make 333 x 3 =~ 1,000,
@cengland0 Thank you. That kind of imprecision through malcommunication really annoys me, too.
@cengland0 @werehatrack My line of reasoning would be if they said 2x smaller, then it means half the original size. 3x smaller would be a third of original, not a third less. Likewise 4x smaller would be a quarter of the original. It wouldn’t make sense for 3x and 4x to have less reduction if the other meaning were taken. True it’s imprecise language, but I think that type of description has been in common use in the past for compression algorithms.
4x and 400% smaller sounds more substantial to some consumers compared to 1/4 and 25% it’s just sales psychology.
KuoH
@cengland0 @kuoh @werehatrack And wouldn’t 1x smaller should mean it is 0?
@cengland0 @kuoh @yakkoTDI Exactly. They can get it right, or we can take whatever meaning we want, and sue their asses into the third world when what they said does not mean what they thought it did.
@cengland0 @werehatrack @yakkoTDI To me, 1x smaller would be the same as 1x larger, it’s a redundant term. Your X factor may be different.
KuoH
@kuoh @werehatrack @yakkoTDI
Not sure they are the same. X factors are different than percentages, right? If I have $1,000 and increase that by 100%, I’ve doubled my money. What X factor do you need on $1,000 to double it? Answer is 2X or two times the money.
What does 1X mean then, I would say that it’s 0% increase. That seems easy math when discussing increases but then what does 1X decrease mean? Is that a 0% decrease so there is no change? If that’s the case then what would 0X decrease mean? The latter sounds more like there was no decrease.
This is why I wish people would stop talking in X factors when discussing decreases. It’s okay when you’re discussing increases because everyone would have a basic understanding that you multiply the original number by the X factor and that’s your increase but it’s not that simple and understood when discussing decreases.
@cengland0 @werehatrack @yakkoTDI As I said before, 1x larger/smaller is a redundant term just as 0x would be and both can express the same net result in this context. We would typically not use the descriptors 1x or 0x but rather “unchanged or the same” amount in these cases. However, our language is full of redundant and inaccurate words and phrases. Some consider that to be the charm of human communications, while to others it’s an irritating artifact. Sometimes it’s better to just go with the flow.
Miscommunications have sparked wars as well as diffused tensions throughout our history depending on a person’s particular viewpoint or ideology.
Just be glad that you’re not communicating, yet, with an all powerful AI or alien being, that is 100% and 1x logically unyielding and accidentally saying that you wish the world’s human population would remain at 0x. Somebody’s got to play Kirk in a universe full of Spocks or it will get to 0K real quick.
KuoH
@cengland0 @kuoh @werehatrack @yakkoTDI
@romellex @chienfou @Chronicle @djslack
Yeah, I agree with that assessment.
And then 3x smaller would be even smaller and then less than 0? Seems reasonable to me.
In the other direction, I’m retired from teaching college math and we math teachers were split almost equally in a discussion in assessing what “two times larger” meant. Two me, it meant “double” added in. So two times larger than 5 would be 5+10 = 15, in other words, two times larger would be an INCREASE by twice as much of the previous and hence three times as big. Others said “two times larger” meant “twice as much” and YES, that IS larger, but then, for them to be consistent, “one time larger” SHOULD be the same size but that is NOT larger and they don’t accept that premise. (And they would argue that was a special case of the usage they argued for.) To me it comes down to whether “x times as big” and “x times larger” mean the same thing. I say not. And that’s the way I was taught back in the day when people paid better attention to established usage. (But nowadays, even younger (under 50?) school teachers cannot get their pronouns correct: “He went with John and I”. NO. Should be “He went with John and me”. I had an English teacher that taught it like this: Leave out the 2nd noun and see if it would be correct. “He went with
John andI” . Wrong. Or “Meand Janewent to lunch.” “Me went to lunch”? Ouch.[And don’t get me started on how adverbs have seemingly disappeared. “Eat healthy!” ? Eat healthy what? Healthy food? As opposed to sick food? If a chicken has been sick, I sure don’t want to eat it. “Drive safe”. How about driving SAFELY? I tend to be a fan of precise language, and you get that when you follow previously established conventions.]
A N Y W A Y, I digress. From the discussion of my original question, I not sure we settled what the expected results of the new format should be, but there seems to be a consensus that the way Google posted it was lacking in precision and generally could be misleading. And that a relative comparison could be better described by a multiplicative percentage of the original, such as the improved file size is 30% of (meaning “times”) the previous. That would be much clearer.
Thanks all. I guess I’ll find numbers when I actually try the various formats.
@phendrick I sort of disageee about the 15 being the correct answer in your example.
If John is twice as old as Bill or John is 2 times as old as bill, would make me think that you take John’s age and multiply it by 2.
As for me, I know I’ve met young people and asked how old they were and I say, “Wow, I’m twice your age.” That doesn’t mean that I’m their age plus 2 more of their age for a total of their age x 3 – it simply means x = their age, y = my age; y = 2x
John: How much money do you have?
Bill: $10
John: Wow, I have twice as much.
How much does John have?
What if John said: Wow, I have 2X your amount.
Then how much money does John have?
Now the problem one: John said, “I have 2X less than you have.”
There’s no way possible to figure out the language to determine how much John has. This is why I hate it so much when people use that X factor nomenclature when discussing less, fewer, or any other decrease. You can make guesses that John means he has $5 but it could also mean that he’s in debt of $10.
I could easily say X factors with decimals and people would get it exactly. I have $100,000 in my IRA but the market was bad and I took a loss of 0.5X. I would assume that means I now have $50,000 in the IRA. But if I said I took a loss of 2X, there’s no way to interpret that without follow-up questions.
@cengland0 Your discussion of John’s age vs. Bill’s, you say
“If John is twice as old as Bill or John is 2 times as old as bill, would make me think that you take John’s age and multiply it by 2.”
I agree with that.
But you are overlooking the distinction I made between being “twice as old as Bill” and that of being "two times older than Bill. That’s the key to everything I posit.
To me, J once older than B would make J = B+B or two B, so two times older than B would make J=B+2B = 3B. If B is 10, then once older than that is 10 plus another 10 and would be 20 and twice older would be 30.
Backing up, should J once older than B hypothetically make J=1*B, that wouldn’t amount to being older than B at all, so has no logic to it at all.
We’re basically debating semantics, and the meaning you attach to something depends on how you learned it, and that can be really hard to get out of one’s system. I argue for the way that I learned them, they made sense then, and still do.
(Might as well argue religion or politics, but I’d rather not.)
I basically wanted to see how others interpreted the statement and my original post, and I have found out. I agree with some, but not others. Thanks for your input.
I know I usually use “k” for thousands when talking money. Like 15k for $15,000. And I know perfectly well in the computer world it is 1024 (2 to the 10th). But I think the impreciseness is OK in casual communication where both sides have the same understanding. “Close enough”. “Slide rule accuracy”.
On the other hand, there are many things where one needs to know with a greater degree of exactness.
I’d guess the person writing the problem quote learned English later in life.
I hope someone in the image-editing field will jump in here and help.
I agree that I would take it as the fraction (3x smaller =1/3 of the TOTAL size, etc.)
Another one that annoys the crap out of me is the use of percentages to indicate changes in numbers that don’t really have a statistical basis. (30% of the group did X. Is the group 3 people? 10? 1000? 100K?). Those smaller sample numbers are frequently then extrapolated to ever larger groups incorrectly.
@chienfou Or how about when they average averages. For example:
2022 - 20%
2021 - 10%
2020 - 30%
Looks like that’s an average of 20% for all years combined, right? 20+10+30=60; 60/3 = 20%
YOU CANNOT DO THAT!!! You must have the raw data from each year that was used to create the original percentages. Percentages are calculated by taking a numerator and dividing it by the denominator. To get the accurate percentage for all years, you must add all three annual numerators and divide that by the sum of all annual denominators.
@cengland0 @chienfou In teaching finite math and introducing statistics there, I used to give my students two questions:
(1) You drive the 180 miles from city A to city B at 90 miles an hour (by yourself) and pick up your mother and drive the return trip at 60 miles an hour (or she’d constantly be complaining). What was your average speed for the driving part of the trip? Choices are
(A) 70 (B) 72 © 75 (D) 78 (E) 80 all in MPH.
Most students would immediately answer with C.
(2) You are taking a class where so far you scored a grade of 90 on a lab practical that counts 20% of the course grade and a 60 on the first major exam that counts 30% of the course grade. What is your current course average based on just these two evaluations?
(A) 70 (B) 72 © 75 (D) 78 (E) 80 %-age points. Most would immediately recognize that © was too high.
Then I’d tell them (B) was the correct answer for both and they were the same problem in disguised form and then I’d discuss “Weighted Averages” and how we were going to run into them in several places in the course. (Expected values, mean values and standard deviations from the mean, Bayesian probabilities, for a few.)
I’d give scores in set X and weights in set Y to match, and have them calculate the weighted average. And then reverse the roles and use X as the weights on Y and discuss the differences in the answers. They were intrigued on how you could arbitrarily weight X by Y or Y by X. (Of course, it made a huge difference in the interpretations of the results.)
Histograms always helped them visualize these.
I’m super glad you’re looking into this. I’ve had almost 20 years of experience in design. It’s always nice to meet someone who’s trying to understand image editing software.
First let’s define some terms.
“Lossy” Means that when you save a file in this format whether it’s a .jpg or a .gif or a .webp file you’re going to loose some information data. Based purely on the file’s compression algorithm. I’d steer clear of these formats if the image content in the file needs to always be the best. Or you’re going to be opening and resaving the content that was previously compressed before. The algorithm just compounds the image content loss as you re-save the files. The compression algorithm of those file formats allows the files to be smaller, but as noted you will lose data.
.tiff .png and a few other’s algorithms don’t degrade the image content nearly as much so they are referred to as “lossless” image file formats.
Next, To help understand the math they are referring to in the description of the the .webp file format they are starting with the base of (for example) a 1000 megabyte [1gb] file in an uncompressed and then saving it as a .PNG file and it may get down to (I’m not going to do the actual math) 300MB .png file. Where as the .webp file format may get it down to 100MB .webp format. Both of which allow for transparency but would you would have to obviously compromise the image integrity if you choose the .webp format. The “3x” term they are using is trying to accomodate for the difference in file size I illustrated above in the 300MB vs 100MB file sizes.
If the comparable PNG file is 120K, I would take that statement to indicate the lossy webp file would be expected to be around 40K.
It’s imprecise, yes, but i take 3x smaller with no extra context to mean 1/3 the size.
Matter of fact if you add context I think I arrive at the same result, just with more math and less intuition. If you say the raw image is 1.2MB, the PNG is 120K, and the webp is 3x smaller, that would be 1/30 the size instead of 1/10 the size, which is the same 40K.