>> Can any of you knowledgable Photoshop users offer a solution to a problem that has occured whilst applying radial blur to an image. The image 10"x7" @ 300ppi, produced in Photoshop 6, shows pixelation or heavy noise in what would appear to be bands within the radial blur. The problem only shows in certain sections of the bands and is smooth in others. Have I missed something here in terms of render quality which is offered with other blur filters but not apparently with radial blur or is this just the way it is and I need to retouch further? <<
Hi Sam, yes this is a well known problem with radial grads. Ensure that dither is on in the grad tool but even when on it does not help as much as with a linear grad. I would add a 50% gray fill layer set to overlay mode, then I would add 2 units of uniform and 1 unit of gaussian noise to the image as a max, try lesser. Perhaps evaluate things at 50% view as well as 100% when judging the noise pattern trade off vs. band correction. Now go to the blending options/layer options for this noise layer. The blend if sliders can be set to only contribute the noise to the gray or colour channel colour ranges that show the noise - by moving the sliders set to gray or channels for the underlying layer (image). The opt/alt key is used to split the sliders. Advanced blending can also be used to disclude say the K channel if working in CMYK so that you are only adding CMY noise (or L in LAB or whatever). This way the layer and sliders can be used to protect highlights or shadows from the introduction of noise. The blend if sliders can be used to target the noise correction to the luminosity values of the bands that you wish to remove, but I think it would be safer just to apply the noise to the entire gradient and only to exclude it from deep shadows/highlights. Anyway - noise is the answer, but only as much as you need to get the job done without mucking up the grad. As well as the grad tool dither and the manual addition of larger filter noise, there is also 8 bpc mode dither noise - but this is too subtle to help and you would have to convert between two spaces which have the same gamma or dotgain so that more banding is not introduced by curving the perfect grad. It would be nice if one of those people with LivePicture could make a quick test between a simple 8 bpc greyscale radial grad in Photoshop vs. a greyscale grad made in LivePicture (which I hear does not suffer banding like this due to using 12 bpc data for alpha masks?). AFAIK you can't paint a gradient in high bit mode in Photoshop - and the only other way I that know to create a 'fake' radial gradation in 16 bpc high bit mode in Photoshop shows exactly the same problems as the 8 bpc version, so I have my doubts if high bits will help, or if they do there must be a reason that Photoshop stinks at radial grads. I will have a quick play and see if rasterizing a vector radial grad (either blend or grad fill) from illustrator will have the same issues. A link to further reading on this topic: http://www.ledet.com/margulis/ACT_postings/ACT-Banding-16%20bit-8-bit.html Regards, Stephen Marsh. =============================================================== GO TO http://www.prodig.org for ~ GUIDELINES ~ un/SUBSCRIBING ~ ITEMS for SALE
