Jump to content

Dazzler

Members
  • Content count

    181
  • Joined

  • Last visited

Posts posted by Dazzler


  1. 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!


  2. 2 minutes ago, firstdefence said:

    Dazzler is basically saying redo the macro again because the dimensions are programmed into the macro. Just uncheck resample before changing the DPI

    No I wasn't saying that at all ... I was testing this myself and finding out there's what looks like a bug here. It's recording the dimensions of an image when it shouldn't be.


  3. @firstdefence Try recording this as a macro. Sure it greys out the size as normal, but it seems to record the image size of the document that I have open when I record the macro. So, say if I have a document at 100px x 100px at 72DPI and I record the macro (unticking the resample button as you've said, and changing the DPI to 300), I end up with my orginal image at 100px x 100px and 300dpi. So far so good. But now I export that macro, and run it on an image that is 200 x 600px @72DPI, and I import the macro and run it, I'd expect to get 200 x 600px @300DPI but instead get 100px x 100px @ 300DPI ?!!


  4. Alternatively, the Select > Select Sampled Color... option when clicked on will give a default selection around most of the items of all colours, however when you then click on a colour on the image it'll sample that colour and select items that are that colour and the tolerance control allows you to expand that range a bit if needs be. You can also keep clicking around in the image area whilst the dialogue is open to select slightly different colours etc.


  5. 1 hour ago, driftinsider said:

    The only way i see now is to buy a Affinity Designer.
     

    Or just get the entire set of Photo , Designer and Publisher :). Seriously though, just drop your cloud subs for a couple of months and there's the money to do just that!

    From what it loooks like you're doing I'd say Designer is a much better tool for that than working in a raster package like Photoshop. Give it a spin and I think you'll see the benefits fairly early on. PS = raster with some vector abilities. Designer = Vector with some raster abilities. Publisher + Designer + Photo = everything in one place using the personas.
     


  6. 1 minute ago, R C-R said:

    So doesn't that at least strongly hint that Move By Whole Pixels is an addition to the Force Pixel Alignment option & not an alternative to it?

    I've always thought of it as an additional constraint, however the two can't really co-exist, so the MBWP takes precedence. Personally I think if you turn on the MBWP it should turn off the FPA button, as that clearly no longer applies.


  7. It happens in other software too. It's because if you take a one pixel line and position it half a pixel to the left (the fractional part of the position) it needs to now render that across two pixels (it will most likely do two half opacity lines instead of one full opacity line), to make it look like that's where it is sat. It's called antialiasing and is the whole reason the 'snap to pixel' style constraints exist.


  8. The problem you have is there's not a lot of area available that shows how it would look without the car. It may be easier to clone a light trail in over the car and continue that up beyong the tree line. It's definitely not an easy job that one as either way you have to deal with lots of gradients because of how the lighting is -  you'll have to find a way to remove the headlights on the road, otherwise there will be an odd bright spot for no reason. You'll probably need some decent painting skills for this!


  9. Affinity Designer 1.7.3.481
    Windows 10 (spec probably irrelevant here)

    So, I'm identifying this as a bug, but there may be other opinions about the way this works. Ultimately, fixing this will have an affect on older designs that use brushes.

     I've made a test brush pattern within designer as a PNG, and exported it (it's 200 pixels height and 568px width). So I then bring this in as a Textured Intensity Brush with a width of 200px to match the height of the exported brush image (black rectangle with white stripes). I then draw a path over the top widthwise of my original brush and assign my new brush - see attached image - brush is in red on the highlighted straight two node path. IMHO I'd expect that brush to be aligned perfectly with the original design now being that the length of the line is the same as the width of the rectangle, and the height is the same. So, heightwise it's fine, but widthwise it's off by a long way. More interestingly is when I use the node tool and drag the end node to the right the pattern will stretch, but then suddenly snap back to allow another repetition to fit in. This then looks very squashed. I would expect the behaviour along a path like this to be consistent - certainly on a straight line, so I think this is highlighting a problem with the maths algorithym - it looks like modulus is bieng used with the repeats but in a way that's not quite right. I'd expect that behaviour to be consistent from the start of the line running to the end, so altering points on the end of a long path with lots of nodes doesn't affect the brush pattern at the start end of the line. It should be working on length of line based from the start node. This is not whats happening, and thus why I'm reporting it as a bug. Brush settings are Width: 200px, size variance: 0%, Opacity Variance: 0%, default pressure curve, Body: Repeat, Head Offset: 0px, Corners: Fold (I tried them all here), Tail Offset: 568px (with a brush graphic width of 568.0px).

    I've just tested a thought I had and discovered that changing the stroke to Butt Cap makes the alignment on the image below perfect, however pulling out the node with that mode on still stretches the brush pattern a lot and then snaps back when there's more than enough space for two patterns to fit. My expectation would be that it would simply add pattern onto the end of the first pattern and allow it to get cut off at the end of the path, rather than stretching one pattern length until there's room for two pattern lengths. It's hard to describe, but it looks to me like the algorythm is working the wrong way round with the modulus. So if the line is 1.5 pattern lengths long, I'd expect it to be using two pattern lengths and cropping the second pattern halfway along. That way the start of the line would always look the same when you adjust the end point.


    image.thumb.png.f57ddafe43238738b652d54be750ffd7.png 


  10. Hmmm this seems to highlight a problem (I'd call it a bug?) with the way the brushes are working. I've made a test brush as a PNG, in a similar way at an exact pixel dimension, and exported it (it's 200 pixels height). So I then bring this in as an intensity brush with a width of 200px to match the height of the exported brush image. Things aren't looking too bad, but with just a two point line you can see a problem - if you grab the node tool and slowly drag one end of the line out to make it longer you'll notice that the brush gets stretched, until it reaches a certain length, then it snaps back to allow another repetition to creep in, which then looks squashed. So rather than just extending the brush pattern and repeating it's actually affecting the brush right back up the path. This seems wrong to me - I think there's an issue with the maths in there somewhere. I can only imagine what a nightmare this is to program, but ultimately the brush should IMHO run consistently from the start point and be aligned with the length of the line, so when working on the end of the path it doesn't affect the beginning of the path. Affecting the brush width should really be the only thing that has an affect on the pattern position along the entire length of the path. Also, if my line is the exact length of my original brush graphic and has a brush width that matches the height of the orginal graphic, I'd expect it to align perfectly - it doesn't even come close.

    Edit: I've reported this as a bug. I have discovered that using Butt Cap on the stroke makes it align perfectly with the original so long as it is the same length, but upon pulling out the end node it still stretches the brush rather than repeating it as I think it should.


  11. I'd also go 8 cores, I have a 6 core PC and just tested doing something in designer whilst looking at the CPU stats and indeed all cores (12 because of the hyperthreading) were being utilised. So with that being the case, in theory with the 6 core you'll get 15.6 - 27 Ghz total, whilst with the 8 core you'll get 18.4 - 38.4 Ghz total. Of course it doesn't work exactly like that due to the way cores are utilised and managed, but as a guide it's probably good enough to go by.


  12. To answer this, there is no minimum, but to explain it better you need to understand how it works. So it's working on the law of averages, whereby if you take a single pixel on the image and it had some noise (random value between 0 and 255 for example, although in reality it would be across three channels and unlikely to hit the extremeties of the value range). For the sake of simplicy let's imagine the orignal image without noise should be the value 127. So if the noise on the original image changes the value to 254. You now need another image with noise at a value of 0 to bring that back in line to where it should be ((254 + 0)/2 = 127). Of course noise isn't likely to be that severe, but it's very unlikely that with 2 images the two noise values would cancel each other out perfectly, but in theory the end result of two images should be closer in truth than either one of the images alone. Obviously, the more images used in the process the more they are likely to smooth each other out. In theory, each noisey value should be biased towards the original 'correct' value, so after adding them all up and dividing the result you should be pretty close to the true value. So yes you can do it with any number of images, but the more the merrier. Two images is unlikely to make a lot of difference, but you may see some improvement. Two should still be better than one. However, there must be a point at which adding more images also becomes fairly pointless. The difference in quality between using two and three images will be greater than the difference in quality between 45 and 46 images.

    The next problem is capturing the images without capturing movement in the image, fine for studio type still shots but not so easy in other situations.


  13. 3 minutes ago, Gigatronix Pete said:

    Thanks for the help here & excuse my ignorance here but how do i "lower the blue channel gamma a bit"?  ...i've never really played with the photo side of things before!

    Any help is much appreciated!

    Add a curves adjustment, then under the histogram there is a dropdown selection for the channel (normally set on Master), switch it to blue. Then, just click and drag on the middle of the line in the histogram and drag down a bit ... it's a live preview so you'll see immediately the effect this has on the image. The histogram is showing the levels in the image ranging from shadows on the left to highlights on the right, and you'll see a peak (or maybe several peaks) where most of your image is sat in that range. You can create several points on the curve and drag them around to affect maybe only the darker areas or create an s-curve to enhance or reduce contrast in a particular tone range. By choosing blue channel you'll only be affecting the levels of the blue, therefore having an effect on the hues in the image, pulling down making it less blue and pushing up increasing the blue and making the image cooler. By using curves you can leave the white area alone, whereas colour balance will have a more general effect.


  14. I don't know whether it's just me having a 'forgetful brain' day or what, but is there a way to move the contents of an individual channel? I have a photo where there is for some reason an obvious offset on one of the colour channels (green I think) and I want to shift it back into position, but for the life of me can't seem to do it! I just need to shift the green channel down a bit. Affine is sort of worknig but way too severe ... 1% is way too far. I've tried typing lots of other values (-0.05 etc) but this seems like a really hard way of doing it. Surely there's some way of just moving by dragging a channel contents around? The defringe tool isn't working here either - I think it's not quite the same thing.


  15. There is a further step you could take to make this more of a 'standard' vector without utilising the 'erase' layer mode (you may find layer modes can be a bit of a problem with certain printers etc, so I normally remove them if possible). You would select all the small curve layers, then from the menu choose Layer > Exand Stroke. Now the strokes will be shapes with a fill rather than single lines with a thickness. You can now use the Add or Subtract functions quite easily to subtact these shapes from the circle (you'll may need to pull the inner layer back out to the top level to be able to combine it with the circle). I tried this, doing an add first of all to join all the little lines into one shape, then selected the circle and the lines layer and chose subtract to remove the lines layer from the circle leaving the shape you wanted without any funny stuff going on anywhere. There was one little hiccup where the end of one of the lines went beyond it's curve that it joined to, but this can be cleaned up easily enough, either choosing a different end cap setting in the stroke options or by simply editing the end result and removing the little nodes in the damaged area. I think fiddling with the stroke end cap options to avoid the overlap in the first place would probably be the best method.


  16. Ok, first of all, looking at your design this should be fairly easy to create using vectors alone, without the need for pixel layers. As always with graphics software there's normally several ways to achieve the same result, so what I say may not be the best or fastest method.

    So first of all, get rid of the pixel mask, that isn't necessary in Designer, as the circle itself can be used to clip the other items by simply dragging the other layer within it (be careful when you drag - note that the blue highlights change position as you drag it over or around another layer, and each of these positions has a different effect when you release). What you want to do is simply drag your curves within the circle so it's just a nested layer within it (so the blue highlight should be showing from the bottom of the purple circle layer with a gap on the left hand side). Once this is placed like that all the inner parts will be clipped by the 'parent' layer - the circle.

    Secondly, get rid of the pixel layer masks within the individual curves - you won't need any pixels here! Once you've done that (you may want to delete the masks one at a time to keep track of what goes where and edit the curves each time), select one of the indivudal curves and switch to the node tool. Where you want the curve to stop and go 'under' the other curves, simply hover over the curve at the intersection of the two curves and you should see a highlighted yellow confirmation that let's you know it's snapping to the intersection. Click to add a point here. Then add another point in the same way where you want the curve to begin again. Then you can select one of these points and choose 'Break Curve' from the context menu at the top of the screen. Then do the same for the other point. (you can actually select both points and do the break curve in one go). You should notice now in the layers panel that your curve has now become three curves. Simply select the one that is the middle of the original curve and delete that layer. Do this for all the remaining curves and you should be good to go, with totally vector, scalable shapes that are very clean and accurate.

    Hope that helps!

     

×

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.