Share the problem, share the code, etc. Read the #📄・posting-guidelines and adjust your post
#need help with a cs50 problem
ashen belfry
dark tusk
Post it here and I shall take a look
gritty lion
gritty lion
Share the problem, share the code, etc. Read the <#1068966762730750105> and adju...
oh wow thats funny. i wasnt sure what you meant bc i was working with bitmaps. idk how i didnt connect the dots
dark tusk
not the jpeg recovery I'm st00pid
gritty lion
youre good bro
dark tusk
So you have some sort of bitmap image and want to apply a blur filter?
gritty lion
yeah and the way to do it is gather all of the surrounding pixels and get an average of the rgb values and put them in the current pixel
dark tusk
Sounds reasonable so far
What does the RGBTRIPLE struct look like? And what's the type of the srnd array?
I assume sum is the count of surrounding pixels you're averaging for the current pixel?
gritty lion
so the rgbtriple struct has three bytes for red green and blue and i made my own struct that has ints so when i add and then divide by sum it doesnt overflow i originally had srnd and avrg as rgbtriple until someone in the cs50 server said it could be overflow
gritty lion
I assume `sum` is the count of surrounding pixels you're averaging for the curre...
yep
dark tusk
Haha I was about to say that my guess is overflowing chars
gritty lion
well you may be right but i have no clue on how to fix it
dark tusk
Change it to use ints instead of chars? Or keep some temporary variable
I think your loop may be off by one as well
Since you fill up the srnd array on indices 0 through 8 (3x3 area), but your loop in line 167 stops at index 7 (m < 8)
gritty lion
oh man that might actually be it.
dark tusk
That looks fishy to me, could be that you're just forgetting about the last surrounding value in some cases
gritty lion
if so how embarrassing
dark tusk
if so how embarrassing
lol off by ones are like the single most common bug in programming
Produced at least two of those today kek
gritty lion
well it actually got two of the check 50 working
dark tusk
So something is still off in there
gritty lion
yeah but thanks for the catch still
ill check and see rq if there is something wrong
dark tusk
Just to confirm, RGBTRIPLE is a struct containing 3 char (1 byte) values?
gritty lion
oh sorry no its an int
dark tusk
Yeah, then that might explain the issue
gritty lion
What did you change the loo pcondition to?
wdym
gritty lion
yeah i did 9
dark tusk
Yeah, then that might explain the issue
You're adding 9 values (index 0-8) of 1 byte each into a variable that's also 1 byte you might be overflowing that variable
(it wraps around and we get a wrong result in the end)
Simple solution would be to introduce 3 new variables of a larger size (e.g. int should be more than enough) above the loop, then we add into those variables, divide them by the count and store them back into the npxl
gritty lion
well for avrg and srnd im using my own struct that they didnt make called rgbtriple int with three ints instead of the rgbtriple with three bytes
dark tusk
So avrg has three ints?
gritty lion
could be but heres something i dont understand
it says it needs 70 85 95 and then it says i have those numbers
dark tusk
Yeah that looks weird 💀
No idea how their test runner works, could it be that they're truncating the output when the issue is in the pixel after?
gritty lion
yeah truthfully i couldnt even guess ill probably just hand it over to people in the cs50 server. youve already helped out a bunch so thanks man i appreciate it
dark tusk
Let's try fixing the rounding first, we can check if the other thing goes away after
gritty lion
alr how would i do that
dark tusk
Do they tell you how to round the values? Seems like you're just rounding down, maybe they expect you to always round up?
gritty lion
i think they just said round. i included a header file math.h and its from a cs50 manual website. its one of the functions on thier website
yeah i just checked the video it said rounded to the nearest integer
i also just checked that function on the website it says it rounds to the nearest integer