Default sketch object, PGraphics vs. PImage

kfrajer

[Q1] Could somebody explain the differences between PGraphics and PImage? Are they related? I believe you can use them similarly but I am not clear when one is preferred than the other.

[Q2] When it comes to something simple as this:

void setup(){
  size(800,400); 
  fill(255)
}
void draw(){
  Background(0);
  ellipse(width*0.25,height*0.25,width*0.5,height*0.5);
}

Is processing creating a PImage object to host the current drawing commands? So when I call the get() function, I get access to this PImage?

Thxs for your comments,
Kf

GoToLoop

[A1] PImage is the parent class: https://Processing.org/reference/PImage.html
While PGraphics is the child class: https://Processing.org/reference/PGraphics.html

As long as the methods & fields from PImage are enough, you don't need PGraphics.

BtW, there are 3rd-party libraries w/ classes which extend PImage too.
For example classes Movie & Capture.

[A2] Sketch's canvas is actually a PGraphics, which is rendered to the PSurface at the end of each draw() cycle. You can access it by calling getGraphics(). :-bd

kfrajer

[Q3] So technically PGraphics have access to PImage's methods as described in the reference here? I believe I have tried before and it complained it was not a PImage object. Could I do casting to convert a PGraphics to a PImage?

[Q4] Now, I can load a PImage and then I can loadPixels to modify color information. Can I draw ellipses and other graphics commands on it?

I know I can do that with PGraphics. However I need to explicitly call beginDraw() and endDraw(). I would say I will always use PGraphics if it wasn't for two details:

1. The point I mentioned in my prev paragraph, not able to access PImage's method like get(), filter(), tint(), etc...
2. Other objects uses PImage, as you @GoToLoop mention above.

In contrast, PGraphics allows to manage transparency slightly different than PImage. I won't get into the details here as this could be another post. I might be wrong on this one anyways.

So far, I have been able to use either object based on my task at hand. Now, I need to decide one over the other.

[Q5] Is there a difference between using the set(), background() and image() methods when drawing a PImage object? I understand from the reference that set() suppose to be faster but to be used only if pixel information is not modified(from the loadPixels() reference). Any other difference?

I am reviewing this link as it could clarify some of my questions: http://processing.github.io/processing-javadocs/core/

Thxs,
Kf

kfrajer

Thxs @GoToLoop for keeping my Q&A format btw :)>-

Kf

GoToLoop

[Q3] & [Q4] Yes, by Java's inheritance rules, a subclass got everything its parent's class got!

However, due to its more complex nature, a PGraphics may need to invoke its beginDraw() method before calling its other methods, even those inherited 1s. =;

And casting doesn't actually convert anything. It simply changes the object's datatype "perception".

For example, if you cast (PImage) over some PGraphics object, that object is still a PGraphics.
Only its members are restrict to those belonging to PImage's class, as long as it's under the cast "effect".

GoToLoop

[Q5] Both background() & set() simply copy the PImage pixels to the PGraphics, disregarding any mode effects and w/o caching the PImage in the PGraphics.

GoToLoop

Now, I need to decide one over the other.

It should mainly be based on whether the chosen datatype got all the members you need to access.

kfrajer

And casting doesn't actually convert anything. It simply changes the object's datatype "perception".

I understand the concept. However when it comes to for example, drawing an ellipse object into the graphics, I do not need to call beginDraw(). In your A2 answer you described that we are working with a PGraphics. When I do

draw(){ellipse,0,0,100,100);}

I am not invoking beginDraw/endDraw. It confuses me. However I did find in the PGraphics reference that this is the option I should be using if drawing an external (alternate) image buffer and that is what I am aiming to.

Kf

GoToLoop

I believe beginDraw() & endDraw() are automatically invoked before & after draw(). :ar!

kfrajer

Related to my last post, if I create a PGraphics object as a secondary buffer to store data, could I load an image into it? This is a theoretical question more than anything else. I was going to try it out except Idk how to load an image into the PGraphics object as:

1. Loading an image requires calling loadImage() with return a PImage (there isn't a function that returns a PGraphics - no that I am aware of)
2. Even if I load an image into a PImage object, how do I transfer the image into the PGraphics object? I guess I will need to call beginDraw()/endDraw()?

This is a theoretical question more than anything. I am thinking I could use it to optimize some image processing code.

Kf

GoToLoop

... could I load an image into it?

We don't load, but "stamp" a PImage onto a PGraphics.
For that we can use: image(), set() or background().

kfrajer

Thanks @GoToLoop!

Kf

Lord_of_the_Galaxy

@GoToLoop you say.

BtW, there are 3rd-party libraries w/ classes which extend PImage too. For example classes Movie & Capture.

So, Movie and Capture are third-party libraries? I thought they were made by the Processing foundation and were "official" libraries.

GoToLoop

Let's just say non-bundled libraries then. :P

Lord_of_the_Galaxy

Better term.