Jump to content


  • Content Count

  • Joined

  • Last visited

Posts posted by Dazzler

  1. 3 minutes ago, R C-R said:

    Look carefully at that screenshot. Note that there is a bright edge at the top of what is circled in red, just like the bright edges for the red, green, & blue channels. If it is nothing more than the overlap, why is it there?

    I see the edges, and yes the combined colour makes a light blueish colour, but if you did as I said and flipped through the channels you'd see there is no fourth value being presented there. The reason for the transparency is so you can see all three channels against each other.

    The histogram is something I've had great difficulty with when teaching people how to use Photoshop. It's not that I don't understand it, it's that it doesn't actually make a lot of sense unless you know what exactly you're looking at. People see is as some sort of brightness curve, but it's not that at all. It's a screen estate measurement. It's really hard to explain becuase really there's not a lot of point in that information apart from to tell you if you have no values happening at the ends which could suggest a lack of contrast in your picture, and as I said previously, not all pictures need a white and a black point. It's not a case of the ends should definitely be at the bottom either. As I said read the y-axis as 'amount of screen taken up with this value' and you see how weird a measurement is actually is. It's easy to calculate from a programming point of view which is why it gets added.

  2. 3 minutes ago, R C-R said:

    There very clearly is a light blue -- just look at the screenshot in the reply above by @walt.farrell & what he circled in red. If it is supposed to be "the overlap/addition of the RGB channels" then why does it dip so low toward the right where red & green peak?

    We need a clear & simple explanation of why it is there, whatever it is, & also if it does not represent luminosity (or luminance) why the histogram does not include one or the other.

    No there isn't a light blue. It is simply the overlay of the colour channels. The dip you are talking about is clearly the blue channel. Just open Photo and grab a photo and flip between the three colour channels in the histogram and you'll see that it is simply a transparency on the three channels and where they overlap you get the combined colour. Nothing more than that. Luminosity is roughly going to be the chart ranging from left to right, not upwards in the histogram as you might expect. ie. A full white image would simply be a spike at the very right hand side of the histogram, that would be full luminosity across the whole image - not a line going across the top of the histogram as you might be expecting.

  3. Here's some things to try if you are wondering quite what the histogram shows.

    1. Fill the entire screen with a flat colour (make a rectangle and choose a fill colour). Now look at the histogram. You'll see three peaks at the RGB values for that colour. Now adjust the contrast quite a bit - what happens to the colour and the histogram?
    2. Fill the entire screen with a black to white gradient in the same way - now look at the histogram. Should be a fairly even line across the top of the histogram.
    3. Place a colour image - then convert to black and white and compare the histograms. Now, on the black and white picture up the contrast to max.  You'll probably notice gaps appearing in the histogram where the values are getting pulling apart from each other. What happens when you do a threshold adjustment? Now flatten the image and gaussian blur the result. How does that look?
    4. Take a photo, look at the histogram and then do a colour balance whilst looking at the histogram and see how it affects it. This may be more apparent if you do it on a black and white image.
    5. Take a few really nicely shot stock photos and place in and look at the histogram. Do the histograms appear as you expected or not?

    I'm not saying there's no point in having a histogram, but you need to know what it's showing you before making decisions based on that info. It's literally counting pixels that have that Red or green or blue value depending on which channel you are looking at and at what point along the graph. Simply doing levels or contrast to pull the values to just inside both ends of the histogram is not always the right thing to do. It's generally a good idea to do this but context is very important. Not all photos should have a white or black point, most should but not all.

  4. 2 minutes ago, R C-R said:

    I think it would be more accurate to say that it shows brightness rather than intensity, but either way the switch to the light blue color makes no sense to me. If anything, it suggests there is some connection to blue levels, while white does not, so why change it?

    Also, why is the background in the dark UI mode black? That just makes it harder than it should be to see the blue channel info. 

    As pointed out by James Ritson, there is no light blue,  it's merely an overlap of the three channels that is producing that colour - the old method of showing white where they overlapped has been removed. As far as I can see the overlap means nothing at all.

  5. 33 minutes ago, R C-R said:

    From https://affinity.help/photo/en-US.lproj/pages/Panels/histogramPanel.html

    So whatever the color, what does it actually show?

    ...and there's the question that's actually kind of hard to answer.

    It's showing you the amount of pixels present at any particular intensity value for that particular channel, where the intensity value ranges from 0 - 255 across the graph (presumeably normalised in some way to fit the graph height nicely).

    Is it that useful? debateable. It can show you potential clipping issues (large spikes at ends of image) - but that's not necessarily always the case (space scene example), or lack of contrast issues  where most of the graph is present in the middle portion of the range (again could be you took a picture of some grey fabric that doesn't have a lot of contrast. It can show you colour casts where there appears to be a stronger presence of one channel against the others - only useful if you know there should be a neutral tone present - grey tones for example.

    Having said that, in general (but not always) it's good to have a nice spread of values across the entire histogram, so that you have some presence in the light areas as well as the dark areas but not falling off the ends. It seems scientific, but it definitely needs to be used with commen sense and an appreciation of what any particular image SHOULD look like in it's representation. The space scene is probably the best example of how this can be mis-leading. Because there is a lot of dark in the picture you'll get a massive spike at the dark end, and a spike at the other end where the stars are represented.

  6. I do think people spend too much time worrying about histograms. They can be really misleading at times .... consider a space scene where you are looking at a bunch stars ... it's going to be a massive peak all at one end of the histogram, and not a lot everywhere else - doesn't mean the photo is wrong or badly exposed - it's just the nature of the shot. If you're looking at a picture of something fairly small on a lawn then the green channel is going to dominate somewhat - doesn't mean the colour balance is out. Relying on a histogram is like putting your camera in full auto mode - a lot of the time it'll give you give you a decent shot but occasionally it'll completely mess up and prevent you from getting a shot that looks right.

  7. I'm not a massive blender fan (Modo is more my bag, and I used 3ds Max before that), but maybe for a newcomer it would make more sense when you don't have a predefined thought pattern for how things 'should' work! It feels laborious in many ways compared to the methods I'm used to, but this is probably just an education thing - I'm just used to working in a set way and I just find myself getting lost with blender really easily - the interface feels a bit alien to me. But then, it is free, and the features it has are very impressive when you consider the fact it is free - it has features that are on par with many commercially available programs. I've seen some spectacular things coming out of it. For the purposes of creating a moon it would be fine - not too much to learn there.

  8. The best way to do this is going to be using 3d software, as it's going to be quite difficult to mimick the way the light catches on the craters, but in the 3d world that becomes fairly trivial, creating a ball with a texture map and bump map - job done!

    Of course, in 2d,  if it's just the overall shadow you want then you could use the above suggestions and maybe paint a few shadows into some of the obvious craters get some extra sense of depth. It's never going to be quite as good or easy as doing it in 3d though. I know 3d software tends to be expensive but there are some available for free that may tackle this easily.

  9. I'm also wondering if I'm missing something ... I'm able to do this fairly easily, unless there's some requirement I'm missing?!

    I've tried to read the original post several times but every time I get to step 3 I'm like 'What?! why?' followed by 'WTF?' at stage 4 and 5.

    So, here's what I've tried and it works fine for me:-

    Type some text. Add stroke (two ways to do this - either use a character style and add the stroke to that which means your text statys 'live' whilst you adjust the stroke, or convert to curves and add stroke after - your choice). You convert to curves after if you used a character style.

    Then using the layers panel, select all the curves within the group that gets created.
    Then Layer > expand stroke.
    Then Press the 'add' boolean.

    Works for me, with all letters whether they are closed or open letters. I end up with a single outlined text shape that includes the stroke width that I put on.

    However, maybe there's some exceptions with a particular font or something? or maybe I'm not understanding the problem correctly?

  10. 40 minutes ago, Old Bruce said:

    I am only curious as to why there is not a clean break from the yellow to the white when starting with  a gradient from 0, 0, 1 to 0, 0, 0. My intuition says there should be a clean break, I am quite often very wrong.

    I imagine this is a dithering algorithm making a blend between two values that won't interpolate further?

  11. Hi Gabe,

    Thanks for looking into this.

    Here's the design file I saved. The black block with the white lines is what I used to make the brush from (just exported that element as PNG), and hopefully the brush line will emain anyway? This raises the question, if I save a file like this does it carry the brush information with it that is needed to recreate the design? If so you should see the red lines over the top, otherwise I guess that may not be showing. I've attached the exported png anyway, and also the brush (hopefully if I've done that bit right!)

    The issue is easily recreatable - it'll be the same with any pattern that has lines running diagonally, or some repeating pattern like this.


    brush test.afdesign



  12. 2 hours ago, zeira said:

    #2 Selection From Layer does not put a selection around the outer edges of the image.  Basically it does nothing visible to the image.

    I need two borders, one internal white border about 1.5mm width all around, then a 4mm black border all around.

    I’ve already found a way to do it without using the very confusing rectangle tool, and as I said it’s more complicated than doing it in Photoshop, and it’s not possible to create a macro for doing it automatically, but it works and I can live with it.

    Here it is:

    1.  Resize the canvas to create an empty border of 1.5mm all around the image.  I couldn’t figure out how to add that one to a macro.

    2. Add a pixel layer and fill it with white.  The whole image will be white, but there is nothing to worry about.

    3. Move this layer to the bottom of the layers stack.

    4. Repeat the procedure to create a 4mm black outer border.


    That's similar to my method, apart from my method is probably a bit quicker as it doesn't involve carefully working out the pixel sizes of the resize document - you just basically make sure it's big enough to allow the extra surround that you want (which you can do easily and visually with the crop tool as I mentioned). Any excess that is left after can easily be removed using the Document > Clip Canvas once you've done everything else. 

    I'm not sure why step 2 didn't work for you? I have the same version and it works for me. You will need to have the image layer selected for it to work with that. Alternatively, you can CTRL(CMD)+click on the thumbnail of the layer to make the same selection. Also, the Select > Outline only seems to give a radius measured in pixels, so the 1.5mm might be troublesome to work out. But it's there as an alternative if you can get it to work. 

  13. I think there's maybe another way to do this. No rectangles and no manually drawing anything, so a fairly accurate way to create set sized frames around a picture.

    1. Open your image and using the crop tool expand it to allow plenty of space for the borders/framing around your image. (pull the top left corner outwards followed by the bottom right corner outwards). Unlock the layer if it is locked.

    2. Choose Select > Selection From Layer. This should put a selection around the outer edges of your image.
    3. Choose Select > Outline. You will now get a dialogue showing a radius control and an alignment option. Choose 'Outside' in the alignment option and then use the radius control to choose the width of the inner frame that you want to create.
    4. Choose Edit > Fill to fill the frame with a colour of your choice.

    5 Repeat steps 2 to 4 for the outer frame (note that step 2 now selects around the edge of the inner frame that you've just created.

    You could continue this to add further frames easily. Hope this helps.

  14. There may be another workaround way for certian situations ... do the normal circular corner tool and choose the largest radius out of the two you want, then bake the corner choose the two nodes at the sides of the curve and select the 'transform mode' icon in the context bar at the top, which you can then use to squash the corner up a bit. Not very accurate but depending on the exact requirements might work in some cases? It's certainly easier and less cumbersome than the boolean method, but is going to be less accurate of course.

  15. 18 minutes ago, GarryP said:

    I still don’t think I’ve managed to get my point across adequately so I’ll try a slightly different tack.
    Please answer this question: “What do you expect the Force Pixel Alignment function to do?”
    Please don’t quote the Help (I can read it for myself) but say what you think it should do from what it’s called.

    IMHO the force pixel alignment should force any creation, movement or adjustment to snap to the nearest pixel.

    However, move by whole pixels is obviously not compatible with that. It's a bit odd and I think the two should actually be completely separate from each other so you can have one or the other turned on, not one as a sub option of the other.

    If you stop and think about it, the Move by whole pixels is actually kind of redundant if you have an option that snaps everything to the nearest pixel anyway. But I guess the thinking is that if you have a shape where not every node is sat on a pixel division and you want to move the shape, you don't want every node on the shape to snap to a pixel division because that would change the shape, so I think this is where the move by whole pixels comes into play? Having said that, that doesn't appear to happen anyway, so the move by whole pixels does seem a bit redundant. It's really handy though if you don't want to snap to pixels and want to move something by whole pixels <stares into space and tries to picture situation that requires this>

  16. Just now, D23 said:

    Oh well, first foray into Photo macros - Result: Fail

    Yeah that's a real shame, and a bit of a surprise. However, in general I really like the way they've separated the 'macro' parts from the image type in Photo - so converting to different formats whilst applying a macro at the same time is a breeze compared to a PS macro, where you have to deal with the 'override file open' type things that never seem to work as you'd expect them to. It's early days with this software - and once they've cracked these early bugs it should become a fantastic piece of software.

  17. Just now, firstdefence said:

    It's not a bug just a limitation of the Macro recording abilities and the available options, if it gave you the cog so that you could just turn of resampling it would be fine.

    Maybe the cog thing is a protected piece of IP that can't be done here? But yeah in PS that would be an option. Anyway, 'tis a bug, has been noted by the devs and hopefully they'll fix it. It should recognise when recording the macro that you've unchecked the resample button, and at that point the dimensions should not be recorded into the macro (because they are not relevant in any way) and it should replicate the DPI change only when played back.

  18. Just now, firstdefence said:

    Was just replying about the apparent inflexibility of this when written as a macro, it's a one shot deal and you'd have to have multiple macros for different dimensions.

    Yeah, that kind of defeats the purpose of recording macros. You should be able to do DPI changes like this without the image dimensions being affected by the macro. I'm saying bug!

  • Create New...

Important Information

Please note the Annual Company Closure section in the Terms of Use. These are the Terms of Use you will be asked to agree to if you join the forum. | Privacy Policy | Guidelines | We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.