8 thoughts on “Beta 5.0 released – Levels control

  • Dwayne

    This will be my last post, but I just wanted to post a couple of examples.

    The darker images are from a month or two ago when I decided the increase in brightness wasn’t worth the loss of details. All were rendered in the Courier New font using the Auto ramp feature and the valid characters “;:. censBHMR”

    http://i14.photobucket.com/albums/a307/Dwayne2005/Gizmo490x4920.gif

    Brightness/contrast seems to only be better for quicker gens.

  • Dwayne

    It’s the best new feature your softwares had so far, IMO; it’s what truly seperates it from ASCGen original. Really impressed. Great work! I hope you can figure out variable width support at some stage. I’m not sure how you might do it. Previously, I suggested you double up chars that are half the size. Rather than that, you might have each of those characters represent a pixel so as to increase quality of the lower images, and the larger characters grab multiple pixels. So, the pixels should be a 1 pixel in width. A 3 pixel character would take 3 of those pixels, averaging out the brigthness. The quality would be sacrificed the larger in width the characters get. (At size 12 Times New Roman, the smallest normal ASCII characters are 3 in width).

    I thought it might be better to set the base value higher, I guess that’s why I’m not the expert. 😀

    I guess you couldn’t really do anything like this before the Auto ramp feature. I can see it being possible now? It’d just be a bit ambitious.

    Anyway, I hope you give it a try! It’d thoroughly surpass any other generator and you’d get people off your back about it! 😉

  • Jonathan

    Basically, my vague idea for variable width is to calculate the values of the characters in that font, then loop through the image checking to see which is the best match at that point and moving along however many pixels it was.

    There would just be a lot of problems integrating it into the program, and I haven’t even tried writing any code for it yet. I really need to figure exactly out how it would work and do a proof of concept test app.

  • Dwayne

    It sounds like a very challenging task that I may have underrated. 🙁 Best of luck, if you should give the test app a go.

  • Dwayne

    I don’t know if this is going to lead anywhere or be of any help, but I’ve just about it and how about this.

    There should be a maximun width setting. I imagine a control that allows you to adjust the max width. When you adjust it, it inserts valid characters into the valid character box. I think it has to sort of centre around this auto generation for the reason that there’s going to due the complex nature of the generation.

    When you adjust the maximun width setting, all the characters that can tally up to that maximun width setting become valid. In the Times New Roman size 12 font, the minimun width for standard ASCII was 3 and the maximun was 15. And combination of characters from 3-15 that tally up to 15 can be included. This way, every combination takes up the exact same amount of space and is sort of localised. The characters that will not be valid would be of width 13 and 14, since there are no width 1 or 2 to tally them up to a total of 15. When you think about it, neither 13 nor 14 width can consistently be integrated in any grid. In Times New Roman size 12, the only characters that correspond with those are % (13) and M (14). They must be excluded in the range of 3-15 widths, otherwise it won’t retain form.

    The program computates all the possible combinations of widths summing 15, from 1 character (15) to two characters (3+12,4+11,5+10,6+9,7+8) to three characters (3+3+9, 3+4+8, 3+5+7, 3+6+6, 4+4+7, 4+5+6, 5+5+5…) to 4 characters (3+3+3+6, 3+3+4+5) and lastly 5 characters (3+3+3+3+3). The entire image will be proportionate at all times.

    The software must then juggle the width and weight aspects… somehow. 😉 Maybe if an adjacent pixel has little variance, it upscales and picks the most respective width combination and weight combination. The more complex combinations, the more detail in the image, so the emphasis should be on having, in this example, 5 character sums rather than 4,3,2 or 1.

    I don’t know, that’s all I could come up with. I have no idea if that’s of any help.

  • Anonymous

    I think there are 4 different possible gens for variable width. The most simplified utilising only the characters of the same width, and you can get okay results from some widths. The second one would be to use factors containing identical widths: 15 and 5+5+5 and 3+3+3+3+3. That would go nicely too. A more complex one is the process I attempted to describe above. Have no idea how that would turn out.

    The last one, the one about looping through the image, is just too complex for me to get my head around.

    I think the key is to get a good grasp of variable width gens first. I think even the first model is better than nothing at all. Then you could expand into the second and third. Or, depending on how ambitious it is, do the 4th concept first or work on it in the long run.

Comments are closed.