• This forum is strictly intended to be used by members of the VS Battles wiki. Please only register if you have an autoconfirmed account there, as otherwise your registration will be rejected. If you have already registered once, do not do so again, and contact Antvasima if you encounter any problems.

    For instructions regarding the exact procedure to sign up to this forum, please click here.
  • We need Patreon donations for this forum to have all of its running costs financially secured.

    Community members who help us out will receive badges that give them several different benefits, including the removal of all advertisements in this forum, but donations from non-members are also extremely appreciated.

    Please click here for further information, or here to directly visit our Patreon donations page.
  • Please click here for information about a large petition to help children in need.

Adding Recommendations to the Calculations Guide: Pixel-Scaling Fix and Rectangle Select Tool Scaling

Status
Not open for further replies.

Flashlight237

VS Battles
Calculation Group
4,144
2,198
Okay, so as it stands, the Calculations Guides provides some general tips to pixel-scaling, however... I feel it missed some important bits. This was a fact I realized after approving a calc, but some images wind up using lines that are very thick, like 5 pixels thick, for pixel-scaling: https://prnt.sc/p0KeXiZjo56R

This may cause issues such as making something seem smaller than it actually is and blocking off important areas for assessment. I drew over the calc'er's pixel-scaling reference to analyze what his line would look like at 5 pixels thick and my preferred 1 pixel thick.: https://prnt.sc/o86vwunvi2dv

As you can see, using Y=413 (https://prnt.sc/Jx85tBJFhMJI ; https://prnt.sc/BvSHsuo5X_Js ) and Y=42 (https://prnt.sc/wVGmgfBgop6f ; https://prnt.sc/O-LE6ZvHjKgg ), the centerpoints used in the image, it is apparent that the five-pixel line is four pixels longer than the 1-pixel line. Of course in that part of the image (where the line is assumed to be 372 px), the difference is only 1.08%, but in another part, the difference is much bigger. At the same image using another reference point (one that is assumed to be 43 pixels long), the 4-pixel difference is much bigger: https://prnt.sc/7FeGnBNdMogx

I'm talking... 9.3%! Now, this isn't exclusive to this one calc. A lot of our calcs did that, and Dalesean even made that mistake with one of his calcs: https://vsbattles.fandom.com/wiki/User_blog:Dalesean027/Two_launches_a_tile_into_the_air

I think it has less to do with the members themselves and more to do with the lack of advisory info as to how to apply pixel-scaling on the wiki's end. See, the way to tell how big a certain line is on an image is by looking at the cursor and the size/shape indicator around the cursor (in GIMP's case, its this dotted-line thing with the pencil tool): https://prnt.sc/aX1yuhEnXnav

Now what Second22 (the calcer whose images I've used for my essay) did do right, however, is add a single pixel to the lines he made. See, in GIMP, the Rectangle Select Tool measures an image directly while the Pencil Tool's lines can be measured with an accuracy of -1 pixel. What this means is that a 20-pixel tall object can be measured at 19 pixels tall with the Pencil Tool.

As such, I think I shall propose additional guidelines to pixel-scaling.

It is recommended that you use a line that is exactly 1 pixel wide when pixel-scaling for the utmost of accuracy. The 1-pixel line will also allow our Calc Group members to get a better look at the events that unfolded in the calc. Depending on the art program used, you may have to add 1 pixel to the measurement you've made in the case of a horizontal or vertical line, or use the Rectangle Select Tool to double-check in the case of a diagonal line. Drawing a right triangle can be used to double-check diagonal lines as well.

In addition, I shall add some statements about Rectangle Select Tool scaling.

Some programs have the Rectangle Select Tool, and in GIMP specifically, it can be used to directly measure objects. When using the Rectangle Select Tool for scaling, you'll want to make the rectangle you used visible enough for people to see, plus you'll need to show the Rectangle Select Tool's options as it shows the size in pixels of the selection. The reason for this is you'll need to screenshot the Rectangle Select Tool in action, as no image program can save the marquee provided in the Rectangle Select tool. An example of a screenshot used for Rectangle Select Tool can be found here: https://prnt.sc/4zOvXsyo0DMQ

The left number is for width (X) and the right number is for height (Y).

For larger images, a 1-pixel line may need to be used in conjunction with the Rectangle Select Tool for validation.

When using the Rectangle Select Tool to measure diagonal lines, you will need to start the selection at the start of the line, then drag it to the very end of the line. From there, you shall use the Pythagorean Theorem and apply both measurements from the Rectangle Select Tool to get the size of the line. For non-linear and non-square objects positioned diagonally, you'll need to draw a 1-pixel line across the object before doing so. In the case you need to measure a different dimension of the same object you measured (ex. an egg), make sure any other measurements needed are properly positioned perpendicularly to the line you just made.

That should make it so that less "thick lines" mistakes are made in calcs. So yeah, there you have it. Hopefully this proposal flies with the wiki.
 
Last edited:
It should be fine for the sake of accuracy, but that would mean I would have to make my lines extra thin since I use Adobe Photoshop Mix on mobile to draw the lines.
 
It should be fine for the sake of accuracy, but that would mean I would have to make my lines extra thin since I use Adobe Photoshop Mix on mobile to draw the lines.
I suppose a recommendation for adding "head-and-tail" bits for lines thicker than a pixel can be done (ex. 1 pixel on each side for a 3-pixel-thick line, 2 pixels thick on each side for a 5-pixel-thick) line, but at the same time, unless you know where your... Dotted-line cursor thing is sitting at, it's finicky.
 
I'm fine with this as well and I generally always have my stuff set to 1px normally aside from my own example above

I can personally say i stick clear of blogs with super thick lines because accuracy goes out the window
 
Last edited:
Seems fine to me aswell. although my calcs do use thicker lines i actually measure with 1px lines and just add additional ones to make them more visible
 
1-pixel lines could lead to a great deal of confusion when actually finding the damn line. Count me as neutral, since this is a trivial affair, but I thought it worth noting that this isn't really all upside.
 
1-pixel works, but generally speaking it's not necessary as long as you don't
a) work with really low amount of pixels
b) chose a line size not much smaller than your measurement target

Like, in most images of drawn characters, even the outlines are larger than 1 px.

And, as Mr. Bambu said, if you have some really HD image people won't even see the line if it's 1px big. In that case, using a slightly larger line can be better.
It's a bit like with the rounding error for computer numbers. It only needs to be small compared to what you calculate.

All in all, I'm not sure if we strictly should instruct to use 1 px.
I think a more flexiable reminder to use reasonably thin line drawing tools so that the length can be validated within the reasonably necessary accuracy might be enough. Maybe with an additional line that for relatively low-res images and small measurement objects 1 px is recommended.
 
Some rule of thumb is always useful.

1 px line is also reasonable.

I think making it a recommendation rather than a rule is a good way to meet the middle.

(I personally draw a very small star or cross to lock the coordinates - I have done this in my prior pixel scaling calcs.)
 
Some rule of thumb is always useful.

1 px line is also reasonable.

I think making it a recommendation rather than a rule is a good way to meet the middle.

(I personally draw a very small star or cross to lock the coordinates - I have done this in my prior pixel scaling calcs.)
I like this one.
 
As usual DontTalk makes sense to me here.
 
How about the following text?
Note that the thickness of the line makes it harder to to tell until where exactly the line goes and, depending on the tool used to measure it, can introduce a certain amount of error in the measurement. Hence it is often a good idea to work with lines that are 1 pixel thick. However, in large or high-resolution scans it can be hard to see such thin lines, making the calculations difficult to evaluate. In such cases one can also use thicker lines, but one should ensure that the thickness is very small when compared to the length of the line, in order to keep the relative error low.
 
How about the following text?
Yeah, I think that makes sense. Though I still think the following would apply:
Depending on the art program used, you may have to add a pixel or so to the measurement you've made in the case of a horizontal or vertical line (depending on the thickness), or use the Rectangle Select Tool to double-check in the case of a diagonal line. Drawing a right triangle can be used to double-check diagonal lines as well.
I'm still firm on adding Rectangle Select Tool scaling guidelines in addition, however.
 
Thank you very much for helping out, DontTalk. 🙏
 
This is extremely dangerous when you get to scans with hundreds to thousands of pixels in 1 direction

A 1280x1940 picture would end up having invisible lines
This is a very good point. 🙏

Should we close this thread then?
 
Probably not. I likely misremembered the purpose of this thread.
 
Probably not. I likely misremembered the purpose of this thread.
The purpose of the thread is mainly to point out the issues with thick lines and the inaccuracies caused by them and ways to solve it, as you can see from the OP. I also put in a suggestion for Rectangle Tool scaling, but that largely got ignored.

I think DT's post here can work along with JasonSith's personal bit of using a little crosshair.:

Note that the thickness of the line makes it harder to to tell until where exactly the line goes and, depending on the tool used to measure it, can introduce a certain amount of error in the measurement. Hence it is often a good idea to work with lines that are 1 pixel thick. However, in large or high-resolution scans it can be hard to see such thin lines, making the calculations difficult to evaluate. In such cases one can also use thicker lines, but one should ensure that the thickness is very small when compared to the length of the line, in order to keep the relative error low.
Some rule of thumb is always useful.

1 px line is also reasonable.

I think making it a recommendation rather than a rule is a good way to meet the middle.

(I personally draw a very small star or cross to lock the coordinates - I have done this in my prior pixel scaling calcs.)
 
I normally use a 3px thick line in Paint.net and I haven't faced any issues regarding inconsistencies and inaccuracies so far.
 
So what is currently left to do here?
 
I believe we just need to add DT and Jason's suggestions, though I took the liberty of rewording it a bit, and you should probably ask again:
It is advisable to use 1 pixel thick lines to minimize measurement error caused by line thickness, although in large or high-resolution scans where thin lines are hard to see, slightly thicker lines can be used as long as the thickness remains significantly smaller than the line length to maintain a low relative error. Sometimes, it can be helpful to add a small crosshair to lock onto coordinates.
Also, rectangle tool scaling has mostly been ignored, so that still requires some talk.
 
Last edited:
Okay, so as it stands, the Calculations Guides provides some general tips to pixel-scaling, however... I feel it missed some important bits. This was a fact I realized after approving a calc, but some images wind up using lines that are very thick, like 5 pixels thick, for pixel-scaling: https://prnt.sc/p0KeXiZjo56R

This may cause issues such as making something seem smaller than it actually is and blocking off important areas for assessment. I drew over the calc'er's pixel-scaling reference to analyze what his line would look like at 5 pixels thick and my preferred 1 pixel thick.: https://prnt.sc/o86vwunvi2dv

As you can see, using Y=413 (https://prnt.sc/Jx85tBJFhMJI ; https://prnt.sc/BvSHsuo5X_Js ) and Y=42 (https://prnt.sc/wVGmgfBgop6f ; https://prnt.sc/O-LE6ZvHjKgg ), the centerpoints used in the image, it is apparent that the five-pixel line is four pixels longer than the 1-pixel line. Of course in that part of the image (where the line is assumed to be 372 px), the difference is only 1.08%, but in another part, the difference is much bigger. At the same image using another reference point (one that is assumed to be 43 pixels long), the 4-pixel difference is much bigger: https://prnt.sc/7FeGnBNdMogx

I'm talking... 9.3%! Now, this isn't exclusive to this one calc. A lot of our calcs did that, and Dalesean even made that mistake with one of his calcs: https://vsbattles.fandom.com/wiki/User_blog:Dalesean027/Two_launches_a_tile_into_the_air

I think it has less to do with the members themselves and more to do with the lack of advisory info as to how to apply pixel-scaling on the wiki's end. See, the way to tell how big a certain line is on an image is by looking at the cursor and the size/shape indicator around the cursor (in GIMP's case, its this dotted-line thing with the pencil tool): https://prnt.sc/aX1yuhEnXnav

Now what Second22 (the calcer whose images I've used for my essay) did do right, however, is add a single pixel to the lines he made. See, in GIMP, the Rectangle Select Tool measures an image directly while the Pencil Tool's lines can be measured with an accuracy of -1 pixel. What this means is that a 20-pixel tall object can be measured at 19 pixels tall with the Pencil Tool.

As such, I think I shall propose additional guidelines to pixel-scaling.

It is recommended that you use a line that is exactly 1 pixel wide when pixel-scaling for the utmost of accuracy. The 1-pixel line will also allow our Calc Group members to get a better look at the events that unfolded in the calc. Depending on the art program used, you may have to add 1 pixel to the measurement you've made in the case of a horizontal or vertical line, or use the Rectangle Select Tool to double-check in the case of a diagonal line. Drawing a right triangle can be used to double-check diagonal lines as well.

In addition, I shall add some statements about Rectangle Select Tool scaling.

Some programs have the Rectangle Select Tool, and in GIMP specifically, it can be used to directly measure objects. When using the Rectangle Select Tool for scaling, you'll want to make the rectangle you used visible enough for people to see, plus you'll need to show the Rectangle Select Tool's options as it shows the size in pixels of the selection. The reason for this is you'll need to screenshot the Rectangle Select Tool in action, as no image program can save the marquee provided in the Rectangle Select tool. An example of a screenshot used for Rectangle Select Tool can be found here: https://prnt.sc/4zOvXsyo0DMQ

The left number is for width (X) and the right number is for height (Y).

For larger images, a 1-pixel line may need to be used in conjunction with the Rectangle Select Tool for validation.

When using the Rectangle Select Tool to measure diagonal lines, you will need to start the selection at the start of the line, then drag it to the very end of the line. From there, you shall use the Pythagorean Theorem and apply both measurements from the Rectangle Select Tool to get the size of the line. For non-linear and non-square objects positioned diagonally, you'll need to draw a 1-pixel line across the object before doing so. In the case you need to measure a different dimension of the same object you measured (ex. an egg), make sure any other measurements needed are properly positioned perpendicularly to the line you just made.

That should make it so that less "thick lines" mistakes are made in calcs. So yeah, there you have it. Hopefully this proposal flies with the wiki.
I agree with the 1 pixel line part.

In terms of measuring diagonal lines with the rectangle tool, I feel like a useful too is Rapidtables Pixel Ruler (It's a simple to use pixel ruler) can be recommended, like how we recommend the angsize calculator. This is just a suggestion however.
 
I'm talking... 9.3%! Now, this isn't exclusive to this one calc. A lot of our calcs did that, and Dalesean even made that mistake with one of his calcs:
I made a thread a while back about how all pixel-scaling is inherently off by 10-20%, and other CGMs said that difference was so small that it doesn't matter.

So now, I don't care too much about differences like that.

Once it gets beyond that level it's an issue, but that requires people to be using lines with a thickness over 10-20% the length of the object, which is patently ridiculous to do in the first place.
As such, I think I shall propose additional guidelines to pixel-scaling.
I disagree, this is too minor to require this much work. Plus, it makes calcs more annoying to evaluate, since for large images it becomes quite difficult to tell where the lines are.
In addition, I shall add some statements about Rectangle Select Tool scaling.
I disagree, this is way overcomplicating things to measure diagonal lines. If we're talking about GIMP, we should recommend its inbuilt measurement tool instead.
How about the following text?
Sure.
I believe we just need to add DT and Jason's suggestions, though I took the liberty of rewording it a bit, and you should probably ask again:
Seems fine.
 
What are the staff conclusions here so far?
 
On line thickness

1 pixel lines: Psychomaster35, Dalesean027, DemiiPowa

This doesn't really matter/has issues: DemonGodMitchAubin, KLOL506, KingTempest, Aguywhodoesthings

Neutral: Mr._Bambu

Flexible recommendation: DontTalkDT, Jasonsith, Flashlight237, Agnaa

On recommendations for diagonal lines

Rectangle Select Tool should be suggested: Flashlight237

Rapidtables Pixel Ruler should be suggested: DemiiPowa

Probably suggest nothing, possibly suggest GIMP's inbuilt measurement tool: Agnaa

I'm surprised that Psycho/Dalesean haven't changed their views, given that even the OP has.
 
Last edited:
On line thickness

1 pixel lines: Psychomaster35, Dalesean027, Aguywhodoesthings, DemiiPowa

This doesn't really matter/has issues: DemonGodMitchAubin, KLOL506, KingTempest

Neutral: Mr._Bambu

Flexible recommendation: DontTalkDT, Jasonsith, Flashlight237, Agnaa

On recommendations for diagonal lines

Rectangle Select Tool should be suggested: Flashlight237

Rapidtables Pixel Ruler should be suggested: DemiiPowa

Probably suggest nothing, possibly suggest GIMP's inbuilt measurement tool: Agnaa

I'm surprised that Psycho/Dalesean/Aguywho haven't changed their views, given that even the OP has.
@Psychomaster35 @Dalesean027 @Aguywhodoesthings @DemiiPowa @Mr._Bambu @DontTalkDT @Jasonsith @Flashlight237

What do you currently think, and are you willing to try to reach agreements here?
 
I would personally use 3-5px for all resolutions barring some really low-res shit (But then again, why the **** would you do that, treat yourself better and get a 4K monitor), and that's on paint.net. There is absolutely no difference in the length of the line within those limits in paint.net.
 
In fact, 1 pixel would be easy to see if you zoom the part big enough.

However, since coordinates would be used anyway, what about the calc maker cite the coordinates of each joints before deducing the length of the line in pixels?
 
Coordinates stop working completely once you start delving into curves or diagonal lines. It only works for lines perfectly horizontal or vertical.
 
In fact, 1 pixel would be easy to see if you zoom the part big enough.

However, since coordinates would be used anyway, what about the calc maker cite the coordinates of each joints before deducing the length of the line in pixels?
That'd add extra work for me, as a user of GIMP. When using shift to draw a line with the paint tool, it shows the co-ordinates of one end of the line, the length of the line, and the angle of the line. I just use that instead of deducing the length from co-ordinates.
 
Status
Not open for further replies.
Back
Top