Using blur techniques in postwork
I'm curious as to what extent folks like to use various blurring or convolving techniques in postwork. I have never previously worried about it all that much because, due to limited computer power, I tended to use comic-esque filters to impart pseudo-realism to my finished work, while "concealing" my shortcomings in terms of being able to produce photoreal Iray work. Now that I have a somewhat more capable lappy, I am able to approach photorealism much more closely, depending upon the quality of the assets I'm using. Still, though, one of the most annoying detractors from CGI art, to my eye, is the razor sharp edges of everything, including characters. Lighting helps to some degree, but sharp edges are the biggest single dead givaway to CGI work, except for perhaps hair. Therefore, I try to use GIMP's softglow, blur, and convolve tools to try to approximate the soft diffusion that our own eyes and occular structures seem to impart to our real world.
How do others approach this issue? GIMP has various types of blur, such as median and gaussian, and softglow, and one can use them in varying strangths, or spot-convolve. But it's ofren a confounding mix to decide what is best, visually.

Comments
Blur is normal, if for no other reason than what DOF imparts; no DOF when the render is trying to be photoreal kills it deader than the dodo. I see some excellent renders posted in threads that don't work becuase of the lack. Other's may have alternative opinion on DOF, but it is something we subconsciously miss as our eyes and brain do it all automatically; everything outside of their focus is blurred to various amounts.
It is also a great way of telling folks where the hell to look, although paying attention to the rule of thirds can help there.
I use DOF as a matter of course, to good effect for the parts of the image that are supposed to be in foreground or background. The problem is that the in-focus parts of the image are still too sharp. Our eyes and brains do not perceive the real world with that degree of acuity. You shouldn't feel as if you could cut yourself on the edge of a character's arm, for instance. I'm wondering how folks like to soften parts of the image that are in focus, DOF notwithstanding.
I am using Bloom combined with Depth of Field. Bloom is in the iRay Render Settings under Filtering (so it really is a postwork filter in reality). Depth of Field is also a postwork filter and set directly on the Camera you are using to render through. After that no post work is needed for me since I'm not trying to simulate motion blur and such things.
I feel it is making me eyes think me renders are more realistic than the sharp edged renders I often do. I have seen sharp edged FHD & higher real digital video and pictures that sharp but are the initial "wow" at the extra detail the sharpness brings it's actually a bit disconcerting to watch such a video.
One of the more common techniques to get a soft focus effect (I use Gimp too, so I'll use the same names found in the most recent version, probably more detailed than you need, but might help others) is to copy the image and past it as a layer over the original. Then use Filters>Enhance>Sharpen (Unsharp Mask)" filter on the new layer. This will sharpen the image a bit more, which is counter intuitive, but what you are really doing here is enhancing the contrast of the details, preparing for the next step. Now copy this layer, and paste it as a new layer. Use Filters>Blur>Gaussian Blur. The key here is that you set the opacity for each layer to your liking (usually between 15%-50%, depending in the results you want. For both filters, you may need to adjust the default filters setting for the best depending on the size/resolution of your image, and the contrast of the original image.
I think there is a GIM'C filter that creates a very similar effect in one step (I don't remember the name), but the plus to doing it this way is that you have a great deal of control over the results, and you have the individual layers that you can tweak more if desired. Hope this helps a little, and you find it useful.
I like bloom too, but I tend not to render with it since you're committed to the effect in the final render. I tend more toward postwork since I can undo that it I change my mind.
Thanks for this! I'll have to give it a try!
Totally agree about the razor's edge outlines. As you say, DOF is essential for the background but the in-focus image is still micro-precision sharp which is not natural. So there's a trade-off to consider. For example, you could increase the Pixel Filter setting to get more blur but this is usually at the expense of some essential detail away from the edges (pores, scars, wrinkles, etc.). I often export the render to a Photo Editor and slightly blur the edges with a blur tool.
On my old lappy, I couldn't use either of your suggestions because of hardware limitations. On the new one, though, they might work. I had to find a workaround for noise because of that computer's limitations. So, to this day, I use Topaz AI to manage what noise my renders may jave in post. That tool is a gem and worth every penny..
This is where I'm finding Topaz Denoise AI to be indispensible. It allows me to render quickly and quite sharply, and then deal with noise in post with granular control. I am finding that GIMP's gaussian blur for controlling sharp in-focus edges is working out best for me among the blurring options.
Yes, I tried the Topaz Gigapixel AI product to enlarge renders. So I would render at half the size I would normally render and then use Topaz to double the size. The results were better than the usual resize options you get with photo editors but still don't bear close inspection - a lot of artefacts. Nevertheless, for a photo-comic story the results are acceptable and the render time is cut down considerably. The product turned out to be a little too expensive for the perceived improvements though.
No, if I don't want bloom I simply turn it off and re-render.
Oh, without nVidia GPU you want to render sharp clear images once and then post work them externally. The other things are if you have a fast enough GPU to experiment with and learn to use the built in features of the iRay renderer and cameras in DAZ Studio. That takes too long to do without nVidia GPU though.
I do the opposite. I render large, scale to about half, thd then use the Topaz product. If my scene is not too geometry-heavy, I get quite clean results for 15-minute renders.
Indeed, I tried it both ways but the Gigapixel AI product doesn't specialise in noise removal. It attempts it but the results are not great. I don't like the IRay denoiser nor the Intel one I tried either.
Do you have any example to post here in the thread?
Would be great to see some images fixed with Topaz AI.