Why Mojo Users Have Not Adopted TG2 (and some solutions)

[efflux]

No Mojo users that I know of except myself have taken up TG2 at least not any who previously did not use TG much. Why? There is a reason for this.

During my use of TG2 I came across three annoying problems. Two of which effect everybody even if they don't get heavily into functions but no doubt these problems will get improved. Ray traced shadow issues with displacements and spiky terrain or large displacements getting cut off at render time, especially at higher quality and scale of render. These problems do also exist in other apps to a lesser or greater degree. There are other things that will get improved in TG2. Render optimizing for example. All things that will improve in the future.

There is however another problem. The big issue in my opinion. The power fractal control or lack of it.

Mojo users have told me that they sent TG2 to the Recycle Bin or Trash (Mac) after testing it. Their words not mine. The first thing they will do when trying TG2 is hook up a power fractal. Then they will look for it's position input (nowhere to be seen). Then they'll wonder where the largest and smallest scale input is. The result scale input, the roughness input etc. Then they may wonder how they can domain distort the fractal. Not possible except with the warper which is very poor.

The general topic was brought up in December last year on this forum and part of my argument was backed up by Calyxa who is not just biased for Mojo since she was the support:

I feel locked out of TG2. The shaders appear to be hard coded and we can't study internal networks. The strength of this upgrade is that we can get at the engine yet we have no info as to how to start using function nodes. In Mojo you just jump in and experiment which is cool. So you learn. You can see how everything works behind the standard UI but not so in TG2. This is a big weak point in my opinion and more info about editing function nodes needs to be supplied or better still some examples.

I agree with a lot of that and am looking forward to seeing more info on nodes and more examples.

Since then I have persevered and discovered that you can do a whole lot with nodes, including recreating more powerful fake stones. The voronoi thread says it all in what can be done beyond what some of the shaders do and I have gone further with altitude driven distorted noise stratas - this looks really cool but no distorted and altitude driven fractal. So all techniques used in Mojo can be used in TG2 except manipulation of the power fractal.

This is a core concept that destroys the chances of tapping into the wealth of knowledge from Mojo users. It was even suggested I write Mojo to TG2 tutorials. I could do this for many things but the power fractal is the sticking point.

The problem lies in the power fractal's lack of inputs to control various functions from other functions. Domain distortion being one powerful technique e.g. using another function to twist the fractal's position around.

I am finding myself using noises rather than fractals for many things and distorting the scales and domains of these with a fractal but not the other way around 

As new basis functions get added to TG2, which I'm sure they will, this situation will get even more frustrating and even with every basis function Mojo has, TG2 would still be crippled.

For those of you who don't know about these techniques it is not so easy to explain. Mojo is the only other program with all this control but it is TG2's main competitor and now a frozen app that will become defunct if not redeveloped. No multi core, no GI, very poor UI, useless Mac version (yet Mac users still stick with it) amongst other things yet most users stick with it or simply don't do any landscape work anymore because TG2 can not supply the power of Mojo fractal control even despite the fact that TG2 has this underlying power. I can understandable this. Without a suitable app I'd prefer just to give up with procedural landscape and do something else until an app that can do it all properly appears. Yes, Mojo has the isosurfaces but that's besides the point here.

Here are some example pictures. The first is a scale distorted (with a power fractal) perlin noise for clouds done in TG2. Not bad. Clouds don't always need all the detail of a fractal:



Then we have the same technique for surface textures (voronoi noise this time). Again, pretty cool and I've used a colour gradient hack to get more than two colours. I now have similar looking strata functions:



Now look at this. It's a Mojo perlin ridge terrain built from a fractal but highly distorted by noise in it's position (the other way around from the above TG2 examples), not distorted in scale but kind of similar shapes are possible and you can distort a noise's position in TG2 if you want. You can use various maths functions in TG2 to get these kind of cliffs instead of a curve graph. However this terrain is impossible in TG2 due to the lack of controlling inputs on the power fractal. In this case no positioning input. You can not distort a fractal in this manner (as far as I know, maybe somebody can tell me otherwise - the warping function is totally limited) despite the fact that TG2 has a capability to do this but the user is stopped from doing it. Distorting a noise with a fractal for terrain (as you can do in TG2) will not give you all the fractal detail that you want:



Other than this issue TG2 is fantastic. Great UI, loads of endless functions can be built, GI, great atmospheric engine etc. There are a few other things that Mojo can do over TG2 but it also works vice versa. Not major issues like this lack of fractal control. TG2 does not have all the difficult to rectify fundamental faults and bugs of Mojo. Mojo users did not voice opinions and so the faults carried on.

[rcallicotte]

I'm glad you're saying something, efflux, and I especially like how you're explaining this so clearly and so graciously.

[Kevin F]

Definitely one for Matt to respond to I think. Hopefully the final version or even the current alpha version addresses some of these issues?
Although I've not seen anything about future versions being able to manipulate the power fractal beyond what we have now.
Kevin

[jo]

Hi efflux,

I have to say I agree with you on a lot of these points. For me the most interesting thing about TG2 is the way it would allow me to create the same sorts of things I did in WaterWorks and such just using function nodes. I'm not so into making zany landscapes like Mojo users, I want to create realistic looking effects for realistic landscapes, but it's all coming from the same sort of place. I have added some simple nodes for functions I'm used to using in making shaders, and which would be familiar to others who learnt to do it the same way as me. I would like to see some fractal nodes which are much more stripped back and allow control much like you'd expect to be able to if you were writing things in code i.e. you provide position inputs, the fractal spits out a value. For sure the fractals in TG2 are powerful but I also would like to see more control over them. One thing I did like about Mojo was that it seemed to match pretty closely what I knew about building shaders. I actually worked on Mojo up until the first Mac release, but I have never really liked it as an application beyond a few things about the approach it took.

I should say that despite being one of the developers of TG2, I really haven't had much time to play with function nodes and such as much as I would like, so perhaps there are ways to do what we want which Matt knows of but which none of us have been able to figure out yet. It would be good if he could fill us in. However, I am going to make it my personal mission to implement a bunch of noise nodes which work a lot more like we would expect, covering all the basic noises. There is a lot to do at the moment, but I'll fit them into what spare time I have. I already have the code to do the noises, it will mainly be a matter of wrapping them up as a TG2 node. To start off they probably won't be a sophisticated as the Mojo ones or the existing ones in TG2. Matt probably won't like them much for various reasons, he has his own take on a lot of these things and in most cases to great effect, but I think it's valuable to have access to a bunch of noises that are well understood and familiar. While they may be older and not so sophisticated they still provide building blocks for more sohisticated effects.

Regards,

Jo

[efflux]

Hi Jo, Thanks for the reply.

Any level of control or any extra noise nodes etc are all useful for anything from realistic to crazy. The power should be centered around the power fractal and I'm finding that it is the opposite. The power is in function nodes which are fully controllable in any way you want leaving the fractal slightly out of the equation which is a weakness so hence my message above. I don't know if anything can be improved with this in future but it leaves TG2 as contender to the procedural throne which is still occupied by Mojo despite all it's lackings and what looks like future death as a viable usable app. Mojo is not an app I really want to use anymore. TG2 is still alpha of course so to be this good at this level is not bad going but the fractal control needs to be better.

At present Mojo is actually capable of much more realistic terrains and rocks and not just because of more basis functions but because of what I describe above. Even although the TG2 renderer and atmospherics are far better. You can get amazing surfaces on your rocks with TG2 but the larger fractal forms are not as good.

[jo]

Hi Efflux,

At present Mojo is actually capable of much more realistic terrains and rocks and not just because of more basis functions but because of what I describe above. Even although the TG2 renderer and atmospherics are far better. You can get amazing surfaces on your rocks with TG2 but the larger fractal forms are not as good.

Well, I think the Alpine Fractal is pretty good. Matt has shown me some stuff done with that aside from a few fractal artefacts is very good. I haven't seen much which comes close with Mojo stuff. I have seen some good stuff, for sure, there is a recent image on Renderosity which is quite impressive. It seems to be few and far between though, possibly because a lot of people seem to concentrate on non-realistic landscapes ( Ken wasn't really interested in realism anyway ).

Regards,

Jo

[efflux]

Mojo started out as a landscape generator to create realistic landscapes but some stuff was thrown in that facilitated weirder stuff. However much of the stranger stuff comes from those very functions that you now see creating very realistic landscapes. Armands has posted a couple recently and ages ago before he stopped for a while. All utilizing Mojo's more realistic capability despite the fact that Armands has created some crazy planets.

The goal needs to be to create an app with maximum procedural control possible to create naturalistic planetary phenomena. This facilitates both earth like realism and anything else you care to create. I'm not really especially into any style although much of my stuff is quite strange. Some of the more extreme was to push TG2,s capability but what I do like is the ability to create something that may be realistic but with at least some slight surreal quality. Probably my stuff will retain this quality but simply get more real looking. This is where lots of power fractal control comes into play.

[Cyber-Angel]

This is what I have been trying to say with various feature requests and other posts I have made here, which have been about one thing and that is providing users with the tools and functionality needed to do what they require weather its some thing wild like a fantasy/ Science Fiction type scene or some thing more realistic. TG2 and future incarnations of the same could be the industry leader in the terrain visualization field and surpass every thing else out there Vue included.

I agree with efflux that more control is required after all shouldn't software be controlled by its user and not the other way around? What we have with the Technology Preview is just the beginning but to quote one of my teachers form Secondary School "Start as you mean to go on" , I realize that as of now that Planetside are a small company, but it is one that is entering the commercial arena now when TG2 is a final commercial product latter this year and that means that it will receive a different kind of attention than it has in the past, with the aforementioned attention comes the responsibility that a commercial product brings with it, and people who will be using TG2 in the professional arena are going to demand a level on control, that right now is not provided or at least is not immediately obvious upon first inspection.

Regards to you.

Cyber-Angel
                   

[jo]

Hi,

Mojo started out as a landscape generator to create realistic landscapes but some stuff was thrown in that facilitated weirder stuff.

I don't think I'd agree with you on that. I was involved with Mojo before it was ever released, and Ken was very definitely all about Mojo being a vehicle for exploring fractal space using landscapes as a way to visualise that, much more so than with the direct aim of creating realistic landscapes, right from the start. Realism was never a priority for him, I don't believe. I've had quite a few discussions about it with him. He seemed to think that realism had been "done" and the real challenge was "hyper-realism" or something, whereas I think most people would probably agree that making visually pleasing cool looking freaky stuff is relatively easy compared to coming up with photorealistic natural scenes. It's been a while, but I certainly remember that was the situation.

The goal needs to be to create an app with maximum procedural control possible to create naturalistic planetary phenomena. This facilitates both earth like realism and anything else you care to create. I'm not really especially into any style although much of my stuff is quite strange. Some of the more extreme was to push TG2,s capability but what I do like is the ability to create something that may be realistic but with at least some slight surreal quality. Probably my stuff will retain this quality but simply get more real looking. This is where lots of power fractal control comes into play.

Well, of course I agree with you, and while for us TG2 is primarily about realism we would also like it to suit those who have leanings toward the more fantastic. We think that a renderer capable of good "Earth like" realism also helps to make the stranger stuff look more convincing. Part of the reason there is a node type interface is it lets ( or should let ) people branch out and create things we might not have concentrated so much on because of our interest in realism. We are not trying to create another Mojo, we are trying to create a photorealistic landscape visualisation app, but there isn't any reason we can't also make it attractive to those who are into what it seems a lot of the Mojo community are.

Regards,

Jo

[jo]

Hi,

I just thought I should add that there are actually a few "primitive" noise functions already, such as the Perlin and Voronoi noise in the Functions->Noise menu. Matt would have to confirm it, but you could probably replicate a lot of the functionality of the Power Fractal shader starting from the Perlin node.

Regards,

Jo

[jo]

Hi Efflux,

Well, I had a bit of a play and came up with this:

[attach=#1]

It doesn't have the same fine ridging your one does, but I think that could be added easily enough by mixing in another fractal or something. It is generated by a Perlin noise node, with the input position distorted by another Perlin noise node. This is the network:

[attach=#2]

It isn't a super clear network, I just bunged a bunch of stuff together. BTW life is much better now with shader previews in separate windows, which is in the latest alpha.

Here's the distorted Perlin noise, prior to the gain and abs functions being applied:

[attach=#3]

This is the noise which is being used as part of the distortion, it's just a regular Perlin noise:

[attach=#4]

I'm not sure if this looks like what you've been getting at. I think while I still agree there needs to be more "primitive" noises like the Perlin noise and Voronoi noise nodes and is important to have good access to their basic parameters ( such as input ), you can maybe do more with what is available now than you think. Don't get too wrapped up in the Power Fractal, as ingenious as it is ( I just had a look at the code, no idea what's going on :-).

Regards,

Jo

[efflux]

OK thanks for replies.

Yes, I have got shapes similar to those Jo but without distorting the power fractal which is my point, that in these circumstances of full control using nodes, the power fractal gets left out of being able to provide the necessary detail that we ideally need.

Ken Musgave was not necessarily into just creating totally realistic landscapes. I realize that but his fractals are the best yet at creating shapes on a larger scale that can get as close as is possible to realistic shapes. This was what he aimed at. My opinion is that his vision of what Mojo should do was totally correct but the failure came in the implementation of certain things and basically bad business moves.

At present Mojoworld is actually capable of creating quite realistic landscapes in the same way that TG2 can do more weird stuff. Mojoworld has one advantage with it's isosurfaces but TG2's displacements are actually more capable than Mojo's standard displacements to get closer to those isosurface forms. TG2 has an edge on some angles like atmospherics and GI which does make for very realistic scenes but once you start to think beyond a stage set then TG2 lacks those vaste and interesting variations needed to create big vistas of varying form. This is not helped by the problems I describe above. Those kind of limitations is why I can't drive the fractal to have those kinds of variations. The forms the fractal actually creates are great. There are no problems with that. Mojo's problems lie is usability more than anything although there are a few things it lacks that an app like this should now have. I harp on about Mojo because it came close to being the dream tool. Tragically it didn't quite get there. What I'd hate to see is TG2 end up with the same fate.

It may be Matt's intention to create a planetary generator that can create things that look like planet earth. This is all well and good because planet earth is the richest planet we know of. The trouble comes in cutting things off from the user which unwittingly limits their capability to actually create realistic forms. I've demonstrated this above. The distorted voronoi noise surface you see has an intricate rich shape. Anyone with any imagination can see that with some different colours then this could form a resemblance to the kind of strata (I actually have a proper altitude driven strata version of this) found in nature but I can't drive the power fractal to achieve these forms. If I could, then I'd have all those beautiful fractal iterations as well. There are numerous other limits I could describe. What if you want the fractal to only follow planet surface position? This can create effects that look like say lichen or something following the downwards flow of water. Another example would be driving the fractals displacement amplitude with another fractal. This can create incredibly varied forms. There are a whole bunch of these limitations with the power fractal at present.

It should now be simply taken for granted that TG2 can achieve very realistic photographic like shots. The Renderosity gallery has a lot of this stuff but the question is, are these pieces of work actually interesting pieces of art? Many of them sadly are not. This is what it must boil down to whether you're creating realistic or fantastic scenes. Non TG2 users who browse these galleries do not care too much about whether the stuff look like photos. This is why those types of work often only get comments from other TG2 users. People know CG software can achieve a good degree of realism. We want this realism to get better, of course or more varied to be precise.

I'm only trying to view my opinions here because of course I want TG2 to be as great as it an be for all users no matter what style of work. I agree with much of what Cyber-Angel says. The more control and the more variety TG2 has then the more it becomes a serious commercial product with wide appeal. It should be a future aim of Planetside to get Vue and Mojo users.

[jo]

Hi Efflux,

Reading over what you were saying originally again, I think what you were getting is basically that you wanted the finer multiscale detail in your example, as opposed to the simpler smoother sort of thing you'd get from the single octave Perlin node, for example. I don't think this is a particularly big deal, essentially it just means there needs to be fractal node like the Perlin and Voronoi ones which is a multi-fractal, perhaps with a controllable number of octaves. Any single multifractal is going to give out on detail at some scale, depending on the number of octaves. Maybe the Power Fractal could be adapted, for that, I'm not really sure.

Anyway, just for laughs and because I don't really play with this stuff as much I should, I did a simple sort of multifractal where I blended together the same fractal as the image I posted above, but at varying scales:

[attach=#1]

Ken Musgave was not necessarily into just creating totally realistic landscapes. I realize that but his fractals are the best yet at creating shapes on a larger scale that can get as close as is possible to realistic shapes.

I don't think that was the case at all, although my experience with Mojo is limited in more recent versions. I think Ken was/is very good at fractals and had a deep understanding of them, and the variety of fractals he was able to add to Mojo meant that realistic landscape forms could be created, but from what I know of Ken, it wasn't really what he was into. TG2 is definitely lacking in the variety of fractals on offer, no argument there.

I am a little bit surprised that there is no way to control things like the input position of the Power Fractal. Hopefully Matt will chime in at some point on why this is, if in fact it is. It is the weekend, I'm only answering today because I'm working Saturday due to taking Friday off to go fishing :-).

It should now be simply taken for granted that TG2 can achieve very realistic photographic like shots.

Well, to be frank, there is still a very long way to go and it can't be taken for granted at all. That's not to say I don't think TG2 is capable of very realistic shots, but it would be a mistake to think we've "solved" photorealism and so can concentrate our limited resources on facilitating things which diverge from an aim of photorealism in general.

The Renderosity gallery has a lot of this stuff but the question is, are these pieces of work actually interesting pieces of art?

From a purely personal perspective, while I'm impressed at the creativity and skill shown in the creation of "unrealistic" type images, they do nothing for me. For me they aren't interesting pieces of art, and I wouldn't hang them on my wall. However, it takes all sorts to make a world and that is simply my personal opinion and not something which influences TG2.

I'm only trying to view my opinions here because of course I want TG2 to be as great as it an be for all users no matter what style of work. I agree with much of what Cyber-Angel says. The more control and the more variety TG2 has then the more it becomes a serious commercial product with wide appeal. It should be a future aim of Planetside to get Vue and Mojo users.

Thanks for very much for the input, please don't think that because of my personal proclivities it isn't appreciated :-). I think it's fair to say we share the same sorts of goals.

Regards,

Jo

[efflux]

Well everyone has different tastes and yes there is loads to go before we have all kinds of different things that truly look photographically real. What I mean is that in some circumstances it's possible to create a scene that really does look like a photo. Then viewers tend to start to take this aspect for granted. They look for something beyond whether it simply looks like a photo or not. What I mean is that TG2 users who have this skill to make a realistic looking scene should consider that the picture also needs to have a feel. Good composition, colours etc etc. This is what I think lacks in the TG2 galleries. For example this picture by Volker is a beautiful image. it's quite realistic but has a lot of artistic content:

http://www.renderosity.com/mod/gallery/media/folder_154/file_1534725.jpg

Another example to describe some of the things I'm on about here is one of mine. Not entirely what I'd like to give as an example because it will be seen as an unrealistic fantasy image but this landscape has some terrain variety at large scale levels but not on on very small scale where you need a fractal. I had to build that up with surfacing. Even if you go into outer space it still shows some large scale variety. Why? because the whole terrain was built from noise functions rather than fractals so I was able to shift it about it more various ways. However it only touches on what I'm talking about and is not as good an example as I'd like to give. Way more is possible even with noise functions. I'd like to show one where I had to use noise functions to create something with more realistic types of features. The reason TG2 suits me slightly better than most Mojo users is because I tend to build low detail terrains anyway and use fractals to provide surface detail but this is not the only way I'd like to work. If you look at some of my more recent images, they are heading in a more detailed route which is why these problems with the power fractal are beginning to bug me.

http://www.renderosity.com/mod/gallery/media/folder_149/file_1489410.jpg

[efflux]

It's probably possible to blend up a load of power fractals to try to get this all over variety whether on a small hill or from space but this is quite difficult to achieve and not so efficent. The easiest way is to use less fractals but have to input control I'm talking about.