[jm82792]

I am doing some tests(this one needs more details and tweaking. It looks a tad melted) that will constitute the first few things I will attempt to make; when I can afford a CNC machine.
Do you think it's a decent piece of work ?

I used GIMP and Blender for it.

[yohudi]

Hi There,

Nice to see what can be acheived in blender using GIMP to create the heightmaps... A good alternative to gmax.. if you are familiar with/ prefer Blender.. You're obviously well on the way to a good grasp of the principles of heightmapping / displacement modelling and their application.

Re the 'melted' look you mention... there are a few things you can do to obtain a 'harder' or more defined look.

First you should make sure that you are viewing an 'un-smoothed' render of the actual mesh surface. Most 3D apps apply render smoothing by default.. based on the fact you need less poygons to render a smooth surface. For surface based work you are best viewing meshes using an unsmoothed render to see the actual mesh surface.

Secondly I would guess that the mesh is generated from a single or combined height map... With this type of map you can create harder lines by keeping a copy of the detail outline on a separate layer when you're creating the heightmap... perform your softening and blurring as usual to get the shaping you want... then when you are happy with the shaping effect, add the detail back in.. on the way, blur and soften it seperately to the shaping areas of the bit map... even if you're using a 'combined' single heightmap you can gain control of detail with this simple adapatation.

Lastly.. I'm not that familiar with Blender, but I have used it's displacement tools enough to know it is quite capable... I think in it's featureset will be the ability to use multiple heightmaps on a single mesh surface... if so this means you now have a great deal of flexibility.

Using this approach, you can, 'stack' the effect of heightmaps on to a mesh surface. By using Image transparancy and a bitmap format such as .PNG (this format retains transparency information an an 'Alpha Map'), separate heightmaps can be restricted to only deform certain areas of the mesh.

I think Blender will have these options... It has to really.. as anything seriously aimed at CGI has to have these kind of features.. Blender might not suit everybody but it's development has at least tried to equal the tools available in high-end CGI apps...

Anyway... I'd say you're well on the way.. a couple of tweaks here and there and you should be knocking them out like nobody's business... well done

TTFN

Danny

[jm82792]

Nice to see what can be acheived in blender using GIMP to create the heightmaps... A good alternative to gmax.. if you are familiar with/ prefer Blender.. You're obviously well on the way to a good grasp of the principles of heightmapping / displacement modelling and their application.

Yeah I have been using Blender for a few years,
and ever since 2.5 came out it's gotten even better.
I'm into materials, texturing, lighting, scenes, compositing, animation and the like as a hobby.

First you should make sure that you are viewing an 'un-smoothed' render of the actual mesh surface. Most 3D apps apply render smoothing by default.. based on the fact you need less poygons to render a smooth surface. For surface based work you are best viewing meshes using an unsmoothed render to see the actual mesh surface.

For most organics you do smooth surface viewing and rendering, however I did forget that most of the time you view it in solid to see the actual geometry, and you do a lot of your modeling that way.
Modeling isn't my strong point so it's no surprise I forget on what I should be doing.

Secondly I would guess that the mesh is generated from a single or combined height map... With this type of map you can create harder lines by keeping a copy of the detail outline on a separate layer when you're creating the heightmap... perform your softening and blurring as usual to get the shaping you want... then when you are happy with the shaping effect, add the detail back in.. on the way, blur and soften it seperately to the shaping areas of the bit map... even if you're using a 'combined' single heightmap you can gain control of detail with this simple adapatation.

That's roughly what I've been doing however I haven't done much of this kind of stuff,
I'll definitely keep what you've said in mind and apply it to what I'm trying to do.

What I did workflow wise:
I took the photo of the flowers, cut the background out, removed and fixed blemishes,
then proceeded to airbrush the soft displacement map on a another layer over the reference.
After that I extracted the shading and the edges using an emboss filter,
and cleaned that up.
Finally I combined everything and tweaked it.
Then it's into Blender, then it's tweaked back in GIMP as Blender's image painting is too slow for some reason.
It's frustrating cycling back and forth between programs.
Due to this frustration I said forget it for this piece and I started sculpting the flowers inside of Blender. I used the displacement map I as my start. I'm hoping I can just bake it down back to a displacement map after I'm done sculpting.

Lastly.. I'm not that familiar with Blender, but I have used it's displacement tools enough to know it is quite capable... I think in it's featureset will be the ability to use multiple heightmaps on a single mesh surface... if so this means you now have a great deal of flexibility.
Using this approach, you can, 'stack' the effect of heightmaps on to a mesh surface. By using Image transparancy and a bitmap format such as .PNG (this format retains transparency information an an 'Alpha Map'), separate heightmaps can be restricted to only deform certain areas of the mesh.

I think Blender will have these options... It has to really.. as anything seriously aimed at CGI has to have these kind of features.. Blender might not suit everybody but it's development has at least tried to equal the tools available in high-end CGI apps...

Anyway... I'd say you're well on the way.. a couple of tweaks here and there and you should be knocking them out like nobody's business... well done

TTFN[/QUOTE]

Thanks for all the information!
What you're mentioning has created a slight conflict in what I should do for a workflow.
Should I do a rough displacement map, then use that as a start to sculpt with inside of Blender(I could even purchase a wacom tablet to make life easier); or (since it's extremely easy) to use multiple displacement maps with multiple amounts of displacement per map-- I could do it all within GIMP, using Blender to preview and mix them.

[yohudi]

What you're mentioning has created a slight conflict in what I should do for a workflow.
Should I do a rough displacement map, then use that as a start to sculpt with inside of Blender(I could even purchase a wacom tablet to make life easier); or (since it's extremely easy) to use multiple displacement maps with multiple amounts of displacement per map-- I could do it all within GIMP, using Blender to preview and mix them.

Hi There,

When it comes to height-mapping there are no hard and fast methods really. Anything that exploits the fundamentals (i.e. dark is low/ light is high/gradient = shape and range of transition) will work. This is the main reason the cnc4free.org ebook stresses and demo's the basic principles before even looking at methods. From your workflow description I take it there's is a fair bit of direct mesh sculpting / smoothing applied. This can be a vaild part of your workflow.. there is no right or wrong... if it works for you then why not ?

The multiple map approach (in chapter six of the cnc4free.org ebook) is a fast, powerful technique. It requires only the basic ability of applying displacement map(s) to a basic mesh form / tweaking a few parameters. It's the 'multi-map' dragon example pictured below. A GIMP .xcf file of the heightmap(s) comes with the support files.

ALL shaping of this mesh is done via the heightmap(s)... Mesh 'edits' = trimming undisplaced faces. Other than edits to bitmap elements in GIMP, there are a few tweaks to gradient curve and displacement amount in the gmax materials editor to blend and position elements at desired levels... that's it... Generally just an extension of basic height-mapping principles and method.

You can probably apply the ebook example maps in blender to work out the necessary blender multi-map method/adaptations. As these maps are proven to work already, you can focus on what you need to do in blender to equate with the gmax 'stacked' maps result...

I wouldn't say that a tablet is required... but it can make graphic design work in general more effective, natural, enjoyable and akin to the 'drawing' process. I have old wacom UD1212 myself.. I also have an el cheapo mini-tablet that came with my last laptop. Both are effective when it come to image-editing. Resolution and tracking on the wacom easily surpass the basic tablet.. but they can both do the job.... If you're planning a long-term addition to your professional's design kit.. invest in a wacom... but if you just want to gain the benefits of tablet over mouse for editing.. a basic tablet will do the job... especially if you don't expect to recoup the cost via work.

hope this helps

Danny

[jm82792]

Hi There,

When it comes to height-mapping there are no hard and fast methods really. Anything that exploits the fundamentals (i.e. dark is low/ light is high/gradient = shape and range of transition) will work. This is the main reason the cnc4free.org ebook stresses and demo's the basic principles before even looking at methods. From your workflow description I take it there's is a fair bit of direct mesh sculpting / smoothing applied. This can be a vaild part of your workflow.. there is no right or wrong... if it works for you then why not ?

I understand what you mean, with art it does seem the ends is what you want and how you get there is up to you.

The multiple map approach (in chapter six of the cnc4free.org ebook) is a fast, powerful technique. It requires only the basic ability of applying displacement map(s) to a basic mesh form / tweaking a few parameters. It's the 'multi-map' dragon example pictured below. A GIMP .xcf file of the heightmap(s) comes with the support files.

Just by looking at your attached image I see what you are doing,
and that methodology makes sense to me.

ALL shaping of this mesh is done via the heightmap(s)... Mesh 'edits' = trimming undisplaced faces. Other than edits to bitmap elements in GIMP, there are a few tweaks to gradient curve and displacement amount in the gmax materials editor to blend and position elements at desired levels... that's it... Generally just an extension of basic height-mapping principles and method.

I see you are limited your poly count.
I can utilize various modifiers to lower the count and they seem to be good enough to keep almost all of the details.

You can probably apply the ebook example maps in blender to work out the necessary blender multi-map method/adaptations. As these maps are proven to work already, you can focus on what you need to do in blender to equate with the gmax 'stacked' maps result...

The thing I love with Blender is that I can utilize non destructive and purely visual modifiers that can be easily applied if needed.
So I can easily have 10 displacement modifiers using 10 different images/layers each with it's own amount of displacement.

I wouldn't say that a tablet is required... but it can make graphic design work in general more effective, natural, enjoyable and akin to the 'drawing' process. I have old wacom UD1212 myself.. I also have an el cheapo mini-tablet that came with my last laptop. Both are effective when it come to image-editing. Resolution and tracking on the wacom easily surpass the basic tablet.. but they can both do the job.... If you're planning a long-term addition to your professional's design kit.. invest in a wacom... but if you just want to gain the benefits of tablet over mouse for editing.. a basic tablet will do the job... especially if you don't expect to recoup the cost via work.

I'm almost 19(budgeted college student), and I've been looking to make some money to justify my hobby of using Blender(and GIMP, etc). I'd like to get good enough to accomplish complex landscapes such as Polulu valley that's 10 minutes from my house.
The primary predicament I have is that I've spent 6 years working with Maya and Blender;
and my hobby's focus has been very broad.
It hasn't been focused on making displacement maps, complex modeling, image painting or working with images.
I can grasp the concepts but lack the experience.

pololu valley - Google Search

[yohudi]

Hi There,

the ends is what you want and how you get there is up to you

Couldn't have put it better myself..

that methodology makes sense to me

Damn right... to anyone with a grasp of 3D geometry and relative mesh displacement those images tell the entire tale

I see you are limited your poly count....I can utilize various modifiers to lower the count and they seem to be good enough to keep almost all of the details.

maximum Polycount in MAX software is restricted by available system resources.. realistically you can get up to about 3 million polys on a system running 2GB RAM. you can do a lot with 3 million polys.. especially if you restrict the mesh to only those faces that will be displaced.

Once you have 'captured' detail using a higher polycount the mesh can be optimized or remeshed to reduce demand on system resources. This process is non-destructive, dynamic and will similarly retain detail. You don't need high polycounts except in areas of high detail.. selective subdivision can be useful. It can also be helpful to separate very high detail elements to separate mesh objects.. common sense stuff really all of it.. especially to anyone with some CGI skills under the belt.

I can utilize non destructive and purely visual modifiers that can be easily applied if needed.
So I can easily have 10 displacement modifiers using 10 different images/layers each with it's own amount of displacement.

yes you certainly can.. and likewise in MAX... you'll find that a great many of the standard or accepted means of performing tasks is pretty common across 3D apps.. after which they then tend to specialise in their particular strength.. 3DSMAX, Softimage and a few lesser known modellers are my own main experience.. with the focus being primarily 3DSMAX.

I'm almost 19(budgeted college student), ....... 6 years working with Maya and Blender..... focus has been very broad....hasn't been focused on making displacement maps, complex modeling, image painting or working with images. I can grasp the concepts but lack the experience.

I would say that, apart from the unbugeted bit, you are in a pretty good position to quickly gain the experience you lack. You have the advantage over many that come to this type of work of grasping the concepts... the rest is practice.

Most CNC based modellers tend to work backwards at least compared to 3D design / CGI heightmapping methods. i.e. They use a function based modeller to generate a heightmap and displace a mesh with it... they just don't usually know it.

They take an image.
Generate a vector.
Split the vector into separate (user defined) closed spline shapes.
Cap the shapes(automatically) to form meshed areas.
Use parameters / adjustments to apply shaping and height.

When adjustments etc are visualised / edited /tuned to requirement heightmaps are calculated (in the background by software) and applied to a plane or other surface. A displaced relief mesh is the result in the viewport.

Apps that work like this are generally based on manipulation of vector (SVG) gradients behind the inteface. The software calculates the gradient values / shape / transitions based on the vector, parameters and edits applied in the view. When smoothing the final output, a bitmap version is generated and a bitmap based smoothing / blurring algorithm is applied followed by displacement of a mesh with it.

These apps hide both the process AND the principles applied in it from the user... detrimentally so in my opinion.. Many CNC users remain oblivious to what they are actually doing when building reliefs with this type of software. They have no grasp of the principles of height-mapping even though they are creating height-mapped reliefs.

When you use the CGI based method you start from the opposite end.. you learn the principles of heightmapping, generate the separate areas from the bitmap. apply gradients and modify them to shape the mesh.. by direct edit, filter or CGI app function.

The underlying principles are identical in both approaches.. except one is FREE requiring a bit of practice.. the other requires 2 to 8 K dollars... similar time learning and refining techniques either way...

As you have a grasp of the principles.. and do not have to spend a lot of money to exploit them, I'd say this gives you good head start... factor in your age and with sufficient motivation there should be no stopping you... seriously.

I attempted my fisrt height-mapped mesh for CNC in about 2007 but it took until about mid 2010 for me to develop/refine the full bells and whistles method. The multi-map method is about 85 - 90 % of the way. The additional 10 - 15 % being use of SVG vectors to generate final heightmaps.

If you focus seriously on adapting the multi-map method for blender then I'd say a maximum of 2 to 3 months regular practice (a mesh a day) and you should be sorted or at least well along the way being an accomplished displacement modeller.. you can then focus on refining/adapting techniques to suit you own area of interest.

Hope this Helps

Danny

[jm82792]

maximum Polycount in MAX software is restricted by available system resources.. realistically you can get up to about 3 million polys on a system running 2GB RAM. you can do a lot with 3 million polys.. especially if you restrict the mesh to only those faces that will be displaced.

Sorry but I cant resist, here is my only CG scene I've ever finished.
It totaled around 10(I figured why not crank it all up for the final) million polygons, although the view-port only handles around 5 million with my hardware.
I used it to learn compositing, as the modeling was easy as it gets.

Once you have 'captured' detail using a higher polycount the mesh can be optimized or remeshed to reduce demand on system resources. This process is non-destructive, dynamic and will similarly retain detail. You don't need high polycounts except in areas of high detail.. selective subdivision can be useful. It can also be helpful to separate very high detail elements to separate mesh objects.. common sense stuff really all of it.. especially to anyone with some CGI skills under the belt.

Yeah I'm thinking you could almost have multiple objects then bake them back together.

yes you certainly can.. and likewise in MAX... you'll find that a great many of the standard or accepted means of performing tasks is pretty common across 3D apps.. after which they then tend to specialise in their particular strength.. 3DSMAX, Softimage and a few lesser known modellers are my own main experience.. with the focus being primarily 3DSMAX.

I've noticed that, for instance Maya has a lot of capacity but it's such a heavy application, it takes forever just to open compared to Blender being so light(although it's missing nice stuff) and with minimal overhead.
Then there is Houdini with it's complex and difficult to learn interface but it puts out amazing special effects and simulations......

I would say that, apart from the unbugeted bit, you are in a pretty good position to quickly gain the experience you lack. You have the advantage over many that come to this type of work of grasping the concepts... the rest is practice.

Thanks!
Hopefully in a few months I'll have some nice stuff put together and I'll get each one milled as tests. After showing every store owner I know, and if the opinions are good; then I'll hopefully start some form of production.

Most CNC based modellers tend to work backwards at least compared to 3D design / CGI heightmapping methods. i.e. They use a function based modeller to generate a heightmap and displace a mesh with it... they just don't usually know it.

They take an image.
Generate a vector.
Split the vector into separate (user defined) closed spline shapes.
Cap the shapes(automatically) to form meshed areas.
Use parameters / adjustments to apply shaping and height.

When adjustments etc are visualised / edited /tuned to requirement heightmaps are calculated (in the background by software) and applied to a plane or other surface. A displaced relief mesh is the result in the viewport.

Apps that work like this are generally based on manipulation of vector (SVG) gradients behind the inteface. The software calculates the gradient values / shape / transitions based on the vector, parameters and edits applied in the view. When smoothing the final output, a bitmap version is generated and a bitmap based smoothing / blurring algorithm is applied followed by displacement of a mesh with it.

These apps hide both the process AND the principles applied in it from the user... detrimentally so in my opinion.. Many CNC users remain oblivious to what they are actually doing when building reliefs with this type of software. They have no grasp of the principles of height-mapping even though they are creating height-mapped reliefs.

Yeah I've noticed(not in your example as I'm totally new to CNC software and such) you can become more or less a monkey pushing buttons if you stay within a certain limit and never extend yourself. I'm still a monkey who pushes buttons
Perhaps if I ever get into scripting I'll understand more of the puzzle.

When you use the CGI based method you start from the opposite end.. you learn the principles of heightmapping, generate the separate areas from the bitmap. apply gradients and modify them to shape the mesh.. by direct edit, filter or CGI app function.

The underlying principles are identical in both approaches.. except one is FREE requiring a bit of practice.. the other requires 2 to 8 K dollars... similar time learning and refining techniques either way...

As you have a grasp of the principles.. and do not have to spend a lot of money to exploit them, I'd say this gives you good head start... factor in your age and with sufficient motivation there should be no stopping you... seriously.

I attempted my fisrt height-mapped mesh for CNC in about 2007 but it took until about mid 2010 for me to develop/refine the full bells and whistles method. The multi-map method is about 85 - 90 % of the way. The additional 10 - 15 % being use of SVG vectors to generate final heightmaps.

If you focus seriously on adapting the multi-map method for blender then I'd say a maximum of 2 to 3 months regular practice (a mesh a day) and you should be sorted or at least well along the way being an accomplished displacement modeller.. you can then focus on refining/adapting techniques to suit you own area of interest.

Thanks I'll continue to learn, then hopefully produce some desirable results and go from there.

[yohudi]

Hi There,

like the spray.. and the volumetric lighting FX..

How much of that was geometry.. and how much composited image.. did you also bumpmap the sea ? ... just wondering where the high polygon usage came in...

Your polygon overhead will be pretty much determined by your subject matter. I don't think you would need to approach anything like 10 million polygons unless you're trying to capture extremely high detail from a megapixel scan or ariel photo...

The mesh for the Chatham Log... the subject of the thread linked below... is pretty detailed and reflects the organic nature of the object.. as visible in the attached image.

http://www.cnczone.com/forums/woodwo...ated_info.html

This mesh was generated by the New Zealand Department of Conservation based on photogrammatic survey of the object in situ.. producing a point-cloud... faces are then constructed by software based on the points... This method applies similar (but not identical) principles as height-mapping.... In height-mapping the XY locations are stipulated by user adjustment to grid dimensions and spacing. Only Z position is derived from an image... photogrammatic methods derive all vertex location data from the hi-res scans.

In conservation work the aim is best 'capture' within the limitations of the technology... Even for a random, organic object like this, the resulting polygon count is just over 1.24 million faces...In height-mapping tests I've found pixel averaging to produce smoother effects if you use a 2:1 ratio between pixel dimensions and mesh subdivision settings..

i.e for a 1,000 x 1,000 pixel bitmap use 500 x 500 segment settings for the mesh surface it maps onto. In the end experimentation with these types of setting will produce the best results on your chosen source material.

Hope this helps

Danny

[jm82792]

Everything you see is mesh minus the volumetric light,
the sea spray was a ton of primitives that I used in a fluid particle simulation and then I vector blurred them. The ocean doesn't use bump mapping, it's all geometry.
Since it's a modifier I cranked the ocean up, way up(probably a stupid idea, but it was a still so why not)
The rocks are around 500K polys.
I couldn't render it all at once, and some of it such as the foam needed (I ran a lot of filters over it) work. I rendered the foam, volumetric lighting(it's a volumetric object, usually used for clouds) and the rest of the scene separately. Then I created a node tree(to work with spec, shadows, AO, etc) and used GIMP to clone(such as the foam, since I needed more of it), touch things up(such as the halo for the buoy) and to fix stuff.

A decent portion of the ocean's detail was lost(plus a lot of it faded in the fog, although I did have the detail decrease with distance).
I have hit the conclusion Blender's rendering engine has a decent amount of limitations compared to say MentalRay or other commercial engines. Although there are some open source ones that are getting there. In the end I wanted photorealism but it ended up more abstract.

Thanks for your input on how high of a polygon count to shoot for,
and the thread, as I enjoyed reading it.

[tracyranson]

I am doing some tests(this one needs more details and tweaking. It looks a tad melted) that will constitute the first few things I will attempt to make; when I can afford a CNC machine.
Do you think it's a decent piece of work ?

I used GIMP and Blender for it.

I really like what you are doing with the relief design. Would you mind providing a copy so that I can work with it in CorelDraw add some detail and see how it machines?

[jm82792]

I really like what you are doing with the relief design. Would you mind providing a copy so that I can work with it in CorelDraw add some detail and see how it machines?

I will happily allow you to use it for your personal use.
I'm still sculpting it, and I'll send you the greyscale displacement map.
I'd also be interested in paying for it to be milled into wood, and having it shipped out here.

I need to figure out how to bake the mesh back to a displacement map, and it's getting late.
I'll get something ready for you in a day or so,
it won't be finished but it will/should be significantly better then what I've shown so far.

[tracyranson]

I will happily allow you to use it for your personal use.
I'm still sculpting it, and I'll send you the greyscale displacement map.
I'd also be interested in paying for it to be milled into wood, and having it shipped out here.

I need to figure out how to bake the mesh back to a displacement map, and it's getting late.
I'll get something ready for you in a day or so,
it won't be finished but it will/should be significantly better then what I've shown so far.

No rush, when are finished send the file to [email protected]

[yohudi]

Hi There,

Thought it might be helpful to post an easy example that clearly shows the methodology and results of 'multi-map' height-mapped relief generation.The aim being to make it easy for any thread reader to see how easy and effective this method is..

These maps were all generated using GIMP gradient functions and filters on areas selected from the 'Detail' outline image.. I think the area selections are easy to see. The results are also posted and there's a zipped 121k face .3DS file, so you can checkout the mesh and do a direct comparison. This is an optimised surface, not the original displaced mesh. I wanted an example that reflected usable mesh density etc....

Almost any displacement modeller that can use multiple maps should produce a similar result based on these heightmaps. All thats required is to apply maps 0 to 5 as explained below.

Map 0: Cuts a shaped area from a plane type mesh - this is optional.. but can help get a 'clean' edge to the mesh..

I generally use plane dimensions approximating image resolution / 10 with segment setting to match. For this example 134 x 187 mm with W segments of 134 and H segments of 187. Further subdivision is added later using gmax.. With other modellers you might need to proportionately increase segment setting at this point.

Sufficient displacement to allow selection and deletion of undisplaced faces is applied.. these faces are then removed Leaving only those faces that will be displaced in the final mesh.

Map 1: The map generates the base form of the mesh. It's quite important as all others displace relative to it.. If it's not right then nothing else can be.

Map 2: 'Boosts' the height of the mesh in the defined area relative to the height of the base form.

Map 3: Ditto

Map 4: Ditto

Map 5: Detail - this 2 colour image evenly 'boosts' displacement in the white areas only. The effect on the mesh is to 'Cut' detail into the surface. Softening of this map using gmax material editor functions softens the edge of the detail.

As you can see from the above.. all the 'work' is in GIMP. Much easier to learn than 3D modelling.. especially a specific type of image editing like this. With a little practice it's not difficult to cultivate an 'eye' for the effects of shading and gradient shape... after that it's largely just a matter of cutting up images to provide the shapes you need and applying suitable gradients and blur.

The secret to success is motivation. Height-mapping is not rocket-science.. It takes a little practice to come to a level where high-quality, predictable results are a matter of course.. It seems only logical to me that it is better to learn the principles that make height-mapping work and apply them using FREE software than to buy into proprietary software (at considerable cost) and spend a similar amount of time learning it and becoming skilled.

From a modeller's perspective, there's the opportunity to reduce your software startup cost to ZERO and maximise the profit margin on each and every mesh you generate. Aqquiring height-mapping skills using FREE software seems a bit of a 'no-brainer' when these aspects are factored in.

Anyway enough opinionation etc... best of luck with your own work.. DO stick at it and I look forward to seeing what I believe will be excellent results in the very near future...

Hope this helps

Danny

[jm82792]

I'm just playing around with the idea of sculpting the mesh within Blender, I will definitely follow your workflow with GIMP(I'm thinking I will use GIMP and not sculpting ).
(Currently I'm dense as a stump when it comes to using GIMP, I know the basics and that's it.)
Here is what I did so far, I know there are issues but I've been playing with sculpting tools(it's almost like a stripped down version of zbrush) and I could see an hour more(since I have more of a bearing of what to do) into this one.
The next thing on the list is an ocean scene with palm trees(I will stick with GIMP), I'll start it soon but I'm currently not having much free time.
Currently my conclusion is that Blender isn't a great idea for sculpting and baking down to a displacement map since I'm not getting the results I'd like(although I'm assuming you can feed a CAM package mesh)
(ignore the quasi HDRI image, I forgot to crop it)

[yohudi]

Hi There,

you're definitely coming on with your mesh scuplting... I like the results... you should keep it as part of your workflow....it will also assist in adding individuality to your work.

If you take a look at chapter 2 of the ebook from my website it explains exactly how GIMP gradients and filters are used to create the type of maps in the previous post. This might help with a scene type model by reducing the sculpting workload.

hope this helps

Danny

[jm82792]

Thanks
Here is what I sculpted it from, before this it was a picture of flowers, all I did was a bit of painting and a filter on the flowers.


It seems case by case it might be decent to use since it seems so easy to mess certain things up when painting. Perhaps it's because I am so bad at drawing. I had my friend ask me if I could see the paper when I drew something, that was while I was part of a class he was teaching. (he did some concept art for Dreamworks)

I'll check out your ebook and read about gradients as I've got an idea of what I'd like but not how to carryt hem out.

Hi There,

you're definitely coming on with your mesh scuplting... I like the results... you should keep it as part of your workflow....it will also assist in adding individuality to your work.

If you take a look at chapter 2 of the ebook from my website it explains exactly how GIMP gradients and filters are used to create the type of maps in the previous post. This might help with a scene type model by reducing the sculpting workload.

hope this helps

Danny

[jm82792]

Would it be terribly difficult to convert this over?
I can easily remove the artifacts such as the house,
or take better pictures since it's by my house.
I really like it because of the clouds, ocean and palm trees,
well maybe more clouds
(this picture was re-sized, I have a much higher res ones)

[yohudi]

Hi There,

I've downloaded both the sources you put up and I'll have a look at them and see if between us we can come up with something you can try in blender, based on your own source. I'm all for expanding the use of FREE techniques and will be a useful insight into the blender side of things, before I complete the update of my ebook....

Re: Source images...

I'm wondering how you intend to 'sell' this work.. I'm getting the impression you're looking at paying to get your output machined, then selling the completed 'piece'... is this right ?

For the Islands type image, you have to address issues of scale / machining requirements from the outset. Accurate reproduction of this type of complex scene demands a strategy that takes machining into consideration.. For very fine detail, toolsize required to clearly machine it is an important factor. This impacts machining times.. and by extension your cost/profit margin.

For woodworking you are pretty much looking at 1 - 1.5 mm ballnose size tools as a minimum... When detail is smaller than this type of tool will cut, a V shaped tool is generally the next option.. using a very small stepover and pass-depth. Alternatively you can adjust scale to increase the detail to a cuttable size. Both strategies have unavoidable impact on machining times. To capture the detail in this image at a cuttable size.. even using V - type tools would mean it had to be a fair size.. and therefore fairly expensive to machine... would this impact the saleability ?

Having seen the type of source you want to use I would suggest NOT modelling it.... I believe you will cut a much better result in much shorter workflow using a direct 'image-carving' method. However, if you're not doing the machining yourself.. the question of how you provide a cuttable source to your machinist would be an issue.

Chapter 8 of the cnc4free.org ebook explains use of the CAMBAM heightmapping plugin to carve directly from images. The plugin is installed automatically in a default install of the FREE beta. It think it will address this type of source allowing the option to use pretty much any toolsize.. including V shaped tools etc.... This will be your best FREE approach to capture such fine detail. The customer would have to accept that fine detail meant smaller tools - longer machine time - higher cost. You will still need to prepare the image somewhat.. as you would for any application...

If I were modelling the island scene I'd go for a stylized / textured version of it rather than an exact reproduction. Don't get me wrong this would still be a pretty credible result.. but it would be an interpretation rather than a reproduction... perhaps this is what you are actually aiming at ??

Hope this helps

Danny

[jm82792]

For that photo I embedded, I understand what you mean by complexity, sometimes I unfortunately get so into something I want to push(more complex, etc)... (perhaps some of it will disappear when I'm not a teen anymore )

It's somewhat devoid of practicality and it would be for more of the heck of it. (something I should avoid :nono: ) Although a stylized approach may make such a concept marketable.


What I'd like to do is make displacement maps that exceed the complexity+quality of the flowers(although if I get them to look good enough they may be worthy), perhaps native fish, a beach house, native plants or an illustration of the now dead sugar cane industry. Once I have around 5 displacement maps that are worthy of being milled(nice looking, good quality, not difficult to mill, etc) I'd like to get a sample of each milled up, and go around various shops to get a rough net figure of what I could sell them for.
If I can actually make a decent net profit even with having them milled and being shipped way out here, I'd like to earn enough money for my own machine. I've always wanted to own CNC machine
If I cannot do anything but break even or lose money then I will have to weigh out purchasing a machine, if one under $2000 is even capable of what I want it to do as I lack knowledge within that realm.

That's basically what I'm currently mulling over; if I can do good(artistically and actually getting it milled/milling it myself with trial and error) work to match what locals and tourists want, plus to make some form of net profit..... even if I get initially minimalistic returns(netting minimum wage for my time) that would hopefully improve with work.

In the end it's somewhat of a battle to make some sort of monetary reward from doing what I like and am passionate about.

[precastguy]

Yohudi,
I have been following your post regarding cnc toolkit and using gmax. I am highly interested in height mapping which will be a great resource in developing tool paths for cnc work. I have been really struggling in learning 'gimp' but slowly getting the hang of it. You posted example of height mapping with the sea horse. My questing if I can work thru your examples of Chapter 2 and Chapter 6 will I be able to generate a clean mesh to create gcode for cnc work? I am figuring it will take about a year before I can build my cnc machine but in the mean time I want to understand everything about height mapping. Thanks so much for all the info you have put out there for us newbies.