9 Replies Latest reply: Sep 6, 2011 6:53 AM by c.pfaffenbichler RSS

    Feature update needed, likely  or possible?

    Astara_ Community Member

      I was editing a drawing and needed to vertically compress, and horizontally stretch  an object that was drawn at about -30 to -40 degrees from horizontal.   I.e. something was too oval shaped and wanted it more circle shaped.

       

      Wouldn't be difficult except that the object (and background) are at an angle).

       

      Of course working on it, I have been using the 'rotate canvas' feature as an aid in keeping my perspective and an aide to help prevent a kinked neck!...

       

      But here's were I think the Photooshop interface either fell short, or could use some enhancement.

       

      Whenever you draw a layer -- especially if you turn on thumbs and clip to the layer boundaries, you notice that no matter the shape of the layer, it puts a rectangle around it.

       

      If you operate with the display rotated, however, all of those rectanges are no longer horizontal with your site, This becomes 'unhelpful' when trying to do anything with the layers...  Including drawing many shapes.  All the layers remain oriented  at their original orientations -- but that a square frame was placed around the object 'horizontally' was a, "somewhat", arbitrary choice to begin with.  Now that I have rotated my canvas, that choice is "off" from how I am currently working. 

       

      I would be nice if when I rotate my canvas, the orientation of the frames change to match -- even though this would likely be a virtual mapping.  It would certainly feel more natural when I go to resize if horizontal was side-to-side and vertical was up-and-down, rather than some skewed angle that is very hard to work with, intuitively.

       

      Am I the only one who thinks this, (or maybe only one who rotates their canvas a fair amount and then tries to manip layers?)

      I.e. this would mostly concern 'Transforms' on objects or layers -- when viewing the picture rotated...

       

       

      Maybe there's an easier way to do what I want?

        • 1. Re: Feature update needed, likely  or possible?
          Noel Carboni Community Member

          It sounds like what you're looking for is what you get with Image - Image Rotation - Arbitrary, except that that causes resampling of all of the raster elements in the document.  You may be able to get around that using smart objects, I suppose...

           

          Just thinking out loud...

           

          -Noel

          • 2. Re: Feature update needed, likely  or possible?
            Astara_ Community Member

            Right -- don't really want to resample 30 or more images just to adjust 1 layer.  I.e. I could rotate the 1 layer, but then it wouldn't be

            against the backdrop of the rest of the pic (composed of layers)...  Of course you can rotate them all, but doing that several times on a 24-bit image isn't gonna help it any...

            • 3. Re: Feature update needed, likely  or possible?
              c.pfaffenbichler Community Member
              but that a square frame was placed around the object 'horizontally' was a, "somewhat", arbitrary choice to begin with.

              I disagree; as the pixels are ordered in a two axis system the horizontal and vertical measurements and bounds carry significance.

               

              But you could always post a Feature Request over at

              http://feedback.photoshop.com/photoshop_family/products/photoshop_family_photoshop

              • 4. Re: Feature update needed, likely  or possible?
                Astara_ Community Member

                I disagree; as the pixels are ordered in a two axis system the horizontal and vertical measurements and bounds carry significance.

                But the frame could have just as easily been placed at a 45 degree angle.   That it was done at the angle it was done at was 'happenstance' by it being the orientation of the object to the base frame when you drew it.   Just as easily, the angle of the frame that composes a layer could have been arbitrarily recorded as being whatever was horrizontal to your screen when you designed it (in the case of have your screen rotated).

                 

                It also need not have been a square, but could have been a circle with the points as polar coordinates rather than cartesian.  That the points were stored as cartesian and the layer was stored as a square were a happenstance of 'ease of calculation'...  Suppose all such

                points were stored as polar relative to each layer, where locking allowed one to lock the center of one later (and it's angle to a fixed point

                and angle in another.  unlocking you could move or rotate a layer w/no loss in image quality.  only upon final rendering to some

                square format like a screen image or print image would such coordinate need to be converted to xy.

                 

                There are other coordinate systems that could have usefulness too, but polar just seemed a natural choice if one wanted to be able to easily rotate and offset layers from each other.  No?  We fell into using rectangular coordinates not because they are 'natural', but rather, just the opposite -- a forced digitization of sections of natural realitiy into small sections that can be displayed on our output devices.

                 

                Ordering more naturally, one might see the vision system as spherical -- and a 2-D representation as ... well, polar.  Rectangular coordinate systems are an accomodation of us to the output devices.   Just as RGB is an accomodation that was recognized as an aspect of the output devices, it was later recognized by Adobe et al, that Hue, Saturation or Intensitity, & Brightness (or Lightness) give a more intuitive way for working with colors.   Coordinate systems just haven't advanced that far yet, as we are easily more used to seeing the world in 'sections'  -- but look at pictures of the globe as they are shifted to 2-D format...doesn't work out so well, sphere's don't map well into rectangles...

                 

                Does my statement make more sense when when explained that way?

                • 5. Re: Feature update needed, likely  or possible?
                  c.pfaffenbichler Community Member
                  There are other coordinate systems that could have usefulness too, but polar just seemed a natural choice if one wanted to be able to easily rotate and offset layers from each other.  No? 

                  In my opinion: No indeed.

                  If for example one were to select part of a layer and move that without rotation the calculational requirements would seem to be much higher for a polar coordinate system and probably also degrade the image worse.

                  But admittedly I’m no expert.

                  • 6. Re: Feature update needed, likely  or possible?
                    Astara_ Community Member

                    You don't think it would be easier for the ****person*** to not have layers at fixed angles that are not adjustable?

                     

                     

                    We are talking usability, not computability.

                     

                    as for image degradation, I don't see why the image would, 'inherently, degrade any worse in a 2-coordinate polar system,

                    than in a rectangular system.

                     

                    One could argue that the math would be more complex, but it depends on what you are doing.

                     

                    If you are rotating or tilting something or mapping onto a 3D curved surface, ... etc, the math might be messier with

                    linear coordinates than curved ones that might more easily map onto a target.

                     

                    I'm sure there'd be more difficult problems for each space, but overall, I'd see them as eventually being

                    identical, though admittedly, I would bet we have alot less familiarity with how one might optimize polar math.


                    It would be a learning curve...BUT...if it could be done in SW --

                     

                    Just like calculating Hue, Brightness/Lightness, Saturation/Intensity, from RGB aren't "natural" for the computer and

                    introduce round off, working with HSB or whatever, is now as naturally offered with no penalty as working wtih RGB -- yet

                    they do require conversion, if you work in HSB, you lose 78.3% of your information!  (RGB has (2.55**3)/3.6 =~ 4.61 times as many color possibilities as HSB).  I don't know how much harm has been caused by such, but I bet the utility of being able to use

                    HSB overcomes any such disadvantage.

                    • 7. Re: Feature update needed, likely  or possible?
                      c.pfaffenbichler Community Member

                      Maybe I’m just misunderstanding your point, but Photoshop is primarily a program for 2D image manipulation and I still think that a two axis sytem of representation makes a lot of sense for that.

                       

                      »as for image degradation, I don't see why the image would, 'inherently, degrade any worse in a 2-coordinate polar system,«

                      Moving a Layer (Edit: or rather parts of a layer) horizontally or vertically in a two axis system means that all resultant pixels can take on the color-values of previously existing pixels.

                      In a radial system that would be quite different.

                      Edit: Maybe sketches can help clarify where I see an issue.

                      rightAngle.jpg

                      radial.jpg

                       

                      In a radial system image »points« might represent different areas whereas in a two-axial system each pixel represents an equal area, which might be benficial computation-wise.

                       

                      How many points should exist per distance unit from the center?

                      Would the number increase with distance?

                       

                      If a Layer would be broadened would the center remain where it was and thus be out of the geometric center or would it be moved, thus necessitating interpolating all previously existing points anew?

                      • 8. Re: Feature update needed, likely  or possible?
                        Astara_ Community Member

                        Well in response the first sentence... Photoshop was a 2D pixel image manip program, but now has font, stuff which I doubt was in the first version, 3D stuff, more vector stuff, ...etc.... i.e. things get added.   I'd really like to see the features of photoshop and illustrator merge, so you wouldn't have to switch tools to use the tools of either, but that's an aside.

                         

                        Your drawings are great!

                         

                        I could see you having problems if you thought of a polar system in that way, but you are taking square pixels and putting them on a polar grid.  Why would the pixels have shape?    I.e why wouldn't they be conceptually points that possess ink that spread out from them, by default, with equal speed/force till they intersect the spread of others.  I.e. if you pick the pixels and move them to another location,

                        they will still have the same shape and distance from each other...

                         

                        It woudl be the same if you you picked up your 'plus' sign on your square system. and moved it in a circle around some arbitrary point such they wer then off kilter from their original placement.  I.e.

                        like blocks.png that ^..

                         

                        If you take the + on the left and only rotated it around the center part way instead of 180 like you have

                        you will end up with the square pixels not lining up on the new pixel grid and would have a simliar

                        failure in accuracy.

                         

                        The idea in a pixel system, on some level is that it's a point.  if you move it, the shape 'auto-fits' into the new location -- i.e.

                        Here-in lies the fallacy promoted by the rectangular system.  The pixel represents some smallest point of representation.

                         

                        In print, it might be little octagons, or similar...looking...like the little dots that spread out -- but each row is offset by 1/2 a sphere from

                        the previous to provide greater density.

                         

                        The polar coordinates might do similar in print, but be simplified in a graphics program.

                         

                         

                        If we had been raised on polar geometry or learned to think how we see -- in radial lines we might have very different expectations of common drawing programs.  But the rectangular system is a way of breaking down the natural world into discrete elements that we can deal with -- it isn't necessarily the system best for representing the real world.  2D doesn't represent 3D real well, and we don't see 3D in a linear XYZ space, but in a spherical space (and not a linear one at that!)...

                         

                         

                        Anyway, wouldn't similar problems affect each coordinate system?

                         

                        To move your points, they would each have a distance and angle from your plane's center (just like x,y points have a distance from a corner).

                         

                        To keep them together, you'd just change the angle to the new location -- there would be no 'boxes' for the points to fit into.

                        • 9. Re: Feature update needed, likely  or possible?
                          c.pfaffenbichler Community Member

                          I think I understand your point better now.

                          And yes, if a point would be just a point and the covered area not defined by a regular pattern of distribution but only be the result of the distance to the nearest points rotation or moving would not seem to be detrimental.

                          But that seems like basically a vector concept which, I suspect, might make very different computational demands.

                           

                          Other issues:

                          In such a scenario what would merging two layers entail? Where would the resultant points be?

                          If points cumulate would there have to be a limit on point-density?

                           

                          What about painterly tool-applications?

                          Would additional points be created (again at the risk of increased density) or existing points’ properties be changed?

                           

                          »I'd really like to see the features of photoshop and illustrator merge«

                          I know you did not say you want the programs merged but only the features, still considering for example Illustrator’s dumb Path-closing-behaviour or the retention of the »Scripts-in-Actions-removed-after-restarting«-bug for at least five versions, I wonder we are not better off if the Photoshop team can keep some distance to the Illustrator team. (Not that I know if or how closely they are cooperating at current.)