This one definitely takes some more liberties with the techniques as explained in the sense I expect people to try it and figure it out from the guidelines I give them but overall I think it produced a really good result
http://gamegurureport.blogspot.com/2016/12/technique-training-bringing-up-level-of.html

A teaser pic:

Thanks... very useful and the other sections

Thank you.

Another interesting read. Do you have these tutorials in a pdf format?

Thanks again for the info! I'm gonna have to put reshade back in, at least until Lee gets the graphics capabilities improved. Happy New Year!

Allways a great read,thank you.

Dave

Nice read, however you should warn people that 4k maps are really soon enough killing the memory allocation -> GameGuru might have a higher memory cap than FPSC, but it has one. I hit it already.

Yeah it gets ugly pretty quick. That's why the previous topic was actually on using smaller maps chained together to keep the memory low.

Well put together again, but I have issues from my own experience with trying this in the past.

You cannot resize an image to twice it's size and add a sharpen effect and call it a higher def picture. It just doesn't add detail, just bloat. So pretty much a waste of memory and of course a good way to slow GG down to a crawl for no real gain. You would need to replace each part of the texture with something that actually does have the extra detail in it. No amount of photo filters will really cut the mustard here as any effect you add to attempt to sharpen the image, will in fact be losing even more details. Not forgetting that the new, hardly different texture, will probably help to slow things down quite a lot, especially if you use a lot of them. Going under 60 fps GG scales down it's resolution, completely negating any difference you may have made. You can disable this of course, but your still stuck with low FPS, probably lower once you disable the scaling.

I have tried this with some models where the texture quality was a little low for my liking in the past. It just doesn't work to any real degree. That's why I always say it's better to go too high with your textures when making them if you can, as it's a simple task to downsize, while up-sizing from my experience, just doesn't work very well. It's not really surprising, as all a computer can do to add detail is to mimic the detail already there.

For the little to nothing you will gain from this graphically, I would say it's not worth the performance hit or the time. Sorry to sound negative here, but I really feel this is a waste of peoples time that could be spent in far more productive areas of their game. If they want higher definition textures than they have. It makes far more sense to make new textures from scratch, or just get an entire model with a higher texture to replace it. That is until we have Bladerunner tech

Of course I only use Paint Shop Pro 7, which is ancient now, and Gimp occasionally, which isn't. Perhaps Photoshop has some awesome tools for this that actually work better. Most people here though aren't going to be using Photoshop because of it's cost. It's a nice idea, but I think most people would be the same as me here, not being able to tell a difference at all, even with the old object and new next to each other directly in front of you. If it's that hard to tell, it's not worth the memory use or performance cost.

Sorry your results weren't as good? I mean I don't know what to say in respect to that.
This is easily the most negative reaction I've had to a post thus far. Not sure how to respond.

My results were good. I clean up the image as best I can using paint.net - the addons you can find for it put it in a class far above gimp and PSP, only slightly less than Photoshop. Your miles were clearly varied.

Good luck on your future work?

Sorry your results weren't as good? I mean I don't know what to say in respect to that.
This is easily the most negative reaction I've had to a post thus far. Not sure how to respond.

My results were good. I clean up the image as best I can using paint.net - the addons you can find for it put it in a class far above gimp and PSP, only slightly less than Photoshop. Your miles were clearly varied. In the case of my choice here I'm obviously not taking a 512 texture to 4096; I'm increasing it to 1024. The beauty of choosing the asylum level was due to it's re-use of textures over and over; because of that the main texture map basically services the entire model set. By improving one, I can increase the quality of a significant number of models at minimum cost.

Hi Bolt Action Gaming, you seem disappointed about the feedback from Dvader, don't be, I've enjoyed your tutorials a lot.

The feedback Dvader has given you is spot on, you could take the information and add to your tutorial e.g. option 1 could work for you if you can't make your own textures or able to source the original (memory could be a problem) or option 2 (Dvader) the way to do it if you have the skills.

I would throw my workflow in the pot as well if it helps anyone, I would create 4096x4096 texture and then reduce it to 2048x2048 or lower this way you keep more detail as you reduce the size.

Keep up the great work

Sorry to be negative. I almost didn't post at all, but I felt that people should know that trying to increase the quality of a texture merely by scaling it up and adding a sharpness filter is really not the way to go. The best way, especially if the UV map is fairly simple like the asylum one, is to completely replace each UV map island with higher quality ones. A far more arduous process I know, but really the only way to effectively do it.

As you say, mileage will vary, simply down to the original textures quality, some will work better than others, but most will only have a marginal improvement, others none at all. The asylum texture is pretty nice overall. Still, I could not see any difference in your before and after image in the tutorial (as I say it's hard to see. When I played with this idea awhile back I really, really tried to convince myself the new bigger image was better).

Sorry to get verbosely miffed in my last post, but as you can see from Emrys post, I'm not the only one to feel this HD method to be a little limited. As he says above don't take it as negative, take it as constructive criticism.

Unless Paint.net IS using Bladerunner tech of course

Actually all this talk of up-scaling is taking me back to a 3DO demo back in the 90's. It had a photo viewer that did a fantastic job of zooming in and clearing up the image. Looked great. Only worked with the photo's provided though and I am guessing they likely had several images for each to get that cool effect.

Something like this you mean.
http://a-sharper-scaling.com

Dave

While I understand your points, I respectfully disagree.
Here's why:
The fact is while a certain amount of reduction is viable (4k to 2k) without a massive loss of visual fidelity, a large amount of reduction (4k to .5k) will result in an enormous loss of quality.

The same applies when trending upwards. Bear in mind, that when using post effect filters, this is MORE pronounced because there's more pixel depth to work with on the actual texture in question. So here:


(without reshade, left side is a seam of the standard res - right side is the higher pixel depth/sharpened texture)


(with reshade, left side is seam of the standard res, right side is the higher pixel depth/sharpened texture)


(left side completely stock texture, stock gameguru - right side reshade HDR level 2 and sharpened texture).

It's important to note...

This is NOT 4k resolution. It's a fraction of a 4k texture. More realistically this exercise was bringing up something from .5k to 1k.

So you're going to see a moderate improvement, but nothing massive. Improving it further requires going even larger. At that point, yes, I'd reskin the model. Now if you don't see a difference, that's fine, but I'm not going to argue the point any further. I made my post, stand by it. I don't make money from these posts; I do my level best to properly inform the community (what's left of it) of things that can be done to hackney this half-built engine into something reasonably useful. There's only so much that can be done to dress up a pig here, fellas. And at this point I'm assuming a certain level of perfectionist that yes, even a slight change from a jagged set of 512 bit depth pixel to 1024 bit depth pixels matters. And I'm that type of person. So I wrote the article in the hopes that others, like me, might benefit. I tend to notice the biggest changes when going two orders higher, say 512k to 2048k. That requires a lot more fiddling and usually a little hand-modification of the texture, which of course this article was supposed to open the door for people to that line of thinking.

If you have more to say on it, by all means. There's a lot of other pieces there that you've flatly disregarded as well - such as changing the plainly pathetic specular mapping to something more functional and realistic. On this point though, I won't discuss any further; I don't feel like we're going to agree or sway the other.