About Pixel Range in version 7.0

[Johnc]

Hi Steve,

Congratulations on the release of MMM version 7.0.
It works well on my 200+ line macro!

I was pretty excited when I saw the Pixel Range new condition and I have just tried it out!

I am assuming that you need to get ALL the pixels to change in the range before it is evaluated true, right?

So I tried it out:

202 | IF | PIXEL RANGE | At Location [X:209 Y:385]::Size [W:3 H:3] | CHANGES::2::30::0 | CONTINUE

Unfortunately it is somewhat a hit and miss and I got stuck sometimes. 

So I switched back to the old trusty Pixel color:

202 | IF | PIXEL COLOR | At Location [X:214 Y:385] | CHANGES::2::30::0 | CONTINUE

and it worked all the time!!

I have to say the color change on the webpage was pretty fast So I am guessing that the single PIXEL COLOR test
was much faster than the PIXEL RANGE test (9 tests in this case), and therefore the PIXEL COLOR was more reliable
in detecting such a fast color change.  Please confirm if my train of thought is correct.

So for higher reliability, I would use pixel color over a large pixel range. 

Another thing is that I cannot think of an application where I need to use pixel range over pixel color.
Seems to me that pixel color would work for me 99% of the time. I am sure others can probably come up with
an application where you must use pixel range instead of pixel color.  

I have a new Condition suggestion though, which should be very easy to tweak from the current code and it should
be much faster and value added to the current pixel range:

The idea is that instead of detecting changes in a large range, just detect changes in 2 pixels farther apart.

e.g.

202 | IF | PIXEL RANGE | At Location [X:209 Y:385]::Displacement [W:3 H:3] | CHANGES::2::30::0 | CONTINUE


It will test pixels (209,385) and (212,388) only. 

202 | IF | PIXEL RANGE | At Location [X:200 Y:400]::Displacement [W:100 H:100] | CHANGES::2::30::0 | CONTINUE


It will test pixels (200,400) and (300,500) only for changes.  This should speed up the condition test and at the same time
test 2 pixels pretty far apart to confirm the webpage/image is indeed fully loaded.

Cheers, John

[Johnc]

Just out of curiosity, I tried the following with a 1x1 Size to see if the speed would improve...

202 | IF | PIXEL RANGE | At Location [X:209 Y:385]::Size [W:1 H:1] | CHANGES::2::30::0 | CONTINUE


Unfortunately the test speed is still slower than the trusty single pixel color test and hence less reliable
than the PIXEL COLOR test.

[Steve]

Hi John,

I'll do my best to answer off your questions:

Q: I am assuming that you need to get ALL the pixels to change in the range before it is evaluated true, right?
A: No. If any pixel in the range changes then it registers as a pixel change and evaluates true.

Comment:

So I tried it out:

202 | IF | PIXEL RANGE | At Location [X:209 Y:385]::Size [W:3 H:3] | CHANGES::2::30::0 | CONTINUE

Unfortunately it is somewhat a hit and miss and I got stuck sometimes.

So I switched back to the old trusty Pixel color:

202 | IF | PIXEL COLOR | At Location [X:214 Y:385] | CHANGES::2::30::0 | CONTINUE

and it worked all the time!!

Reply: The location of Pixel Range X and Y refers to the top left most pixel start. So your example X 209 and Y 385 with a Width of 3 and Height of 3 covers pixels 209,210,211 and 385,386,387



Your following Pixel Color condition starts at X 214 and Y 385. That location is outside of the previous Pixel Range condition.

Q:

I have to say the color change on the webpage was pretty fast So I am guessing that the single PIXEL COLOR test
was much faster than the PIXEL RANGE test (9 tests in this case), and therefore the PIXEL COLOR was more reliable
in detecting such a fast color change. Please confirm if my train of thought is correct.

A: Pixel Color is faster at run time than Pixel Range because Pixel Range is detecting a range of Pixels and also internally the code is a little more complex for Pixel Range and takes longer to complete.  Saying that though the time difference between the two conditions at run time would be less than a few milliseconds.

Comment:

Another thing is that I cannot think of an application where I need to use pixel range over pixel color.
Seems to me that pixel color would work for me 99% of the time. I am sure others can probably come up with
an application where you must use pixel range instead of pixel color.

Reply: The main reason to choose Pixel Range over Pixel Color is to monitor a range of pixels for a change. A use case may be that the change in pixel location is slightly random or unpredictable. For this you could cover the range where the pixel is likely to change and still detect a change. Doing this with a Pixel Color condition would require a lot of Pixel Color conditions whereas Pixel Range covers the range with one statement.

Comment:

I have a new Condition suggestion though, which should be very easy to tweak from the current code and it should
be much faster and value added to the current pixel range:

The idea is that instead of detecting changes in a large range, just detect changes in 2 pixels farther apart.

e.g.

202 | IF | PIXEL RANGE | At Location [X:209 Y:385]::Displacement [W:3 H:3] | CHANGES::2::30::0 | CONTINUE


It will test pixels (209,385) and (212,388) only.

202 | IF | PIXEL RANGE | At Location [X:200 Y:400]::Displacement [W:100 H:100] | CHANGES::2::30::0 | CONTINUE


It will test pixels (200,400) and (300,500) only for changes. This should speed up the condition test and at the same time
test 2 pixels pretty far apart to confirm the webpage/image is indeed fully loaded.

Reply: Hmmm yeah i can see a use case for that. Cheers for the suggestions.



I hope this covers a few things on Pixel Range (www.turnssoft.com/conditions.html#pixelrange) and helps clarify some things for you.

Cheers,
Steve.

[Johnc]

Nov 1, 2018 4:34:27 GMT Steve said:

Q: I am assuming that you need to get ALL the pixels to change in the range before it is evaluated true, right?
A: No. If any pixel in the range changes then it registers as a pixel change and evaluates true.

oic... any pixel!! An "OR" condition. I can use it to detect mouse-over! Good going Steve!

202 | IF | PIXEL RANGE | At Location [X:200 Y:400]::Displacement [W:100 H:100] | CHANGES::2::30::0 | CONTINUE





It will test pixels (200,400) and (300,500) only for changes. This should speed up the condition test and at the same time

test 2 pixels pretty far apart to confirm the webpage/image is indeed fully loaded.

In my suggestion here, I think both "OR" and  "AND" conditions are useful depending on the situation.

So I switched back to the old trusty Pixel color:



202 | IF | PIXEL COLOR | At Location [X:214 Y:385] | CHANGES::2::30::0 | CONTINUE



and it worked all the time!!

My apology... I tried the following again:

202 | IF | PIXEL COLOR | At Location [X:209 Y:385] | CHANGES::2::30::0 | CONTINUE


with similar reliable results.

Thanks for all the crystal clear explanations Steve! It helps a lot and I have got plenty of new toys
to play with now! Thanks a lot! 

Cheers,
John

[asemmohsen]

helloo
i was wondering after using pixel range 


202 | IF | PIXEL RANGE | At Location [X:209 Y:385]::Size [W:50 H:50] | CHANGES::1::0::0 | CONTINUE


how would i be able to make my mouse move to the exact location of where the pixel changed ?

[Steve]

Hi asemmohsen

You can't do that with the current version unfortunately but a heap of pixel and image features are coming to version 8.3 soon.
Stay tuned.

Regards,
Steve.

[Johnc]

Hi Steve,

Really looking forward to the new pixel functions coming in v8.3!

1. New Action - PIXEL CAPTURE

2. New Action - MOUSE TO IMAGE
...

Cheers, John