Cicero D'Avila

I have continued in my efforts to make a good print of the sculpture inspired by the Cicero D'Avila pose in my previous post.

Firstly, I remade the arm pose, tipping the woman's head back to improve the angle of the face and prevent the arm support material from falling across it.

 

This smaller version was printed on a delta printer in red PLA and showed that the model was printable although the underside of the arms is still very ragged following removal of the support material.

At this point I decided to change the arm position again, opting for a more vertical configuration that would hopefully avoid the need for any support material. 

Once again I decided to shortcut the process of making the model printable by using Project Miller to re-skin the mesh. Unfortunately, the model still had some errors which resulted in malformed Gcode when I tried to slice it with Slic3r and Kisslicer. This time I found that the new version of Cura (I'm using 13.12) offered some very useful advanced options including Combine Everything (Type B) under the Expert settings. The Gcode looked very good in Repetier Host so I set off the 12 hour print and this time I finally managed to produce a model that does justice to the sculpture I had planned.

Here's a timelapse video made inside the printer using Octoprint.

Some detail shots taken before the smoothing process.

And after smoothing in acetone vapour to remove the print lines and support blemishes.

One more video of the finished product.

Body Language

I have just completed one of my best projects so far. It was inspired by another great Bruno Birkhofer photo called Body Language.

I took special care with the pose modelling to capture the lines as faithfully as possible, carefully placing the limbs as close as possible to each other without overlapping. I reused a short hair style of my own creation.

Here is a screenshot of my version in Blender.

The next important step was deciding how to print it and what kind of support to generate. I used Kisslicer to generate the printer files again, because although a new version of Slic3r has just come out with supposedly much improved support generation, I haven't had time to test it yet.

I decided to to make one change from the settings I used for the previous Birkhofer print and that was so go up from Medium to Dense for the support layer, just to see what would happen.

The other genius idea I had was to turn the print vertically for printing, which changed the position and amount of support material required. If I had printed the model exactly as it appears in the picture above, all of the chest and head would have need support underneath it. By rotating it to a much more vertical angle almost none was needed in these areas.

I was so impressed by the new settings that decided to make another video of the support removal process. You can see how my preferred tool is a flat bladed screwdriver which breaks the weak bonds between model and support with less risk of damage to the delicate areas. The support was certainly more dense than before and, in my  view, held the model better whilst still coming away cleanly.

Support removal video

Before and after smoothing

Finished print

Recline further

I decided to persevere with Recline and try to fix some of the problems.

Following an invitation from Sublime to try his calculator (see the comments in Recline), I learned that my setup could work better at 0.2 mm layer height if I increased the extrusion width from 0.4 to 0.48 mm. That should certainly help to improve layer bonding somewhat.

I also reasoned that the delamination in the first print had been caused by excessive part cooling with the fan. I know, I know. ABS isn't supposed to need a cooling fan but I still believe that with steep overhangs it just prints a bit better with one; I just had it turned up a bit too high.
The next problem to work on was the fingers. I started another couple of prints and both times the fingers failed again. Something different was going on here.

I thought about this for a long time and eventually came up with three ideas for possible solutions.
Firstly, the way that I had splayed the fingers out made the overhang angle more than 45 degrees. Changing the pose slightly to make the fingers more vertical should help.
Secondly, I had changed the support gap setting in Kisslicer from 0.5 mm to 1 mm to try and make it easier to break away; I suspected that this was actually making the support less supportive, so I put it back to 0.5 mm.
Finally, it dawned on me that the only thing supporting the fingers was the weak interface layer attaching them to the raft. This hadn't been a problem with Curl because there were relatively large areas attached to the raft but this print was quite literally clinging on by its fingertips, and in the end the lateral drag of hot plastic from the rapidly moving nozzle was proving too great for the weak bonds. The solution for this print was to abandon the raft and find a better way to get the fingertips to stick to the bed. 

One option I considered was to print the model on a plinth. This would provide the strongest and most reliable base but wasn't really in keeping with the rest of the series so I decided to keep it in reserve. 
Eventually I resorted to my favourite method which has worked well in the past. I went back into Blender, opened the model in point edit mode and used 'Box select' to highlight all the points in the bottom millimeter of the model and then used Scale Z 0 (zero) to flatten the bottom of the model and finally moved all the selected points up a little to the most natural position.
A liberal splash of ABS juice on the printbed and I was ready to have another go.

This time it went much better. No serious delamination at all and the fingers are looking much more solid. The left thumb appears to have shifted and reconnected at the base, but this should be relatively easy to repair.

Now I just need some time to do the support removal and acetone vapour bath.

Recline

Here's my new project.  I've started a collection of potential sculptures on Pinterest and, having printed several curled up models, I fancied trying a more open, reclining pose like this one.  I have now discovered that this is one of the 'Square Nudes' series by Carsten Witte.

The amount of support material required should be fairly minimal although the steep angle under the upper back may need some.  Additionally, the palms of the hands are raised off the ground and will also need supporting. 
Another of the challenges will be to deal with the area where the thighs and calves will inevitably overlap after posing the model.

More to follow ...

Turns out that I still haven't quite got this 3D printing lark completely sussed yet. The modelling went quite well. I even broke with tradition and tried leaving the mouth open a bit to allow the front teeth to show. This, in turn, meant having to devise a different technique for closing off the open edges in the mouth area.

As usual, I opted to use Kisslicer to generate the gcode because I still find that it produces the best support material. The problems started when I began to print. In spite of the fact that I was using the same settings as I have for all my other prints (ABS at 240C, 0.2 mm layer height, 0.4 mm line width), the extrusion seemed thinner than usual and the inter-layer bonding was poor. The model finally appeared from the build chamber with several large cracks including a complete break in one of the arms.

The other major problem was that the fingers appeared to be poorly formed and looking as though they might come away with the support material. I haven't tried removing the support yet, due to lack of time, but I really don't hold out much hope for saving this print. 

One of the factors that has changed is my new hot end which has a 0.4 mm nozzle, whereas my previous one had a 0.5 mm diameter hole. Slic3r includes a place to enter the diameter of your nozzle but apparently Kisslicer does not.  I have never quite understood the logic of this, so I don't really know whether it could be a possible factor in explaining the latest failure.

I'll probably give it another try and maybe just turn up the flow rate a bit.

Curl

I decided to have a go at reproducing the pose in Jian Xu's picture shown at the bottom of my last post. Tricky parts included getting the fingers into the correct position, particularly since it seemed to push the right elbow into odd places, and dealing with the deep mesh overlap where calves met thighs and thighs met trunk.

I'm still struggling to get good quality solid hair and I don't always like the models looking bald! This time I used Sculptris to fashion a short hair style starting from a simple sphere. It's not great but it kind of works.

Here is the cleaned up version in Blender beside the reference picture.

Metamorphosis looked beautiful in the translucent blue but, being made of PLA, I couldn't use the acetone vapor bath to smooth the surface. PLA isn't soluble in acetone. I knew that I wanted to have a smooth surface on Curl, so I thought I would try printing a copy in black ABS.

My first attempt was sliced in Slic3r 0.98 but again I was disappointed with the way it generates support structures, so I abandoned that print and sliced again in Kisslicer. This version looked much more promising and I have learned that the initial raft is an indispensable feature for supporting this type of model.

Here is the model on the print bed and in close up, showing the support structures (under LED lighting).

Here is the bottom section of the support material after it was peeled away from the model. I am still impressed with how well the support separates from the model given that they are both made of the same material.

And here is the final version after processing in acetone vapor.

Metamorphosis

This weekend I created a new sculpture based on the photo of Edward Watson performing in Kafka's The Metamorphosis. I tried to reproduce the pose as accurately as possible and was very pleased with the result.

 I had hoped to print the sculpture without turning on the automatic support feature in the slicing software, but as soon as I had finished posing the model it was clear that this would not be possible. As well as the shallow slopes on the thigh and arm, there were points, on the tip of the chin and the underside of the breasts, that had nothing beneath them. 

I used Kisslicer to turn the STL model into printer instructions and enabled medium support. This time I decided to try printing at 150 micron resolution to reduce the visibility of the layers. I also wanted to find out how well the support material could be removed from a PLA print so I used a translucent blue filament for the 12 hour print.

The support material was more difficult to remove from PLA than with the grey ABS, but after a bit of work it came away surprisingly cleanly.

Here is a quick video of the model on a turntable to show it from all angles.

http://youtu.be/MRIWulQLC5U

And here are a few photos.

Sascha

I love this striking photo by Sascha Hüttenhain and I wanted to use this as the inspiration for a new sculpture.

I had to use a bit of artistic licence to create a pose that would capture elements of the photo but remain printable on my 3D printer.

Here is the digital version of the pose that I wanted to try and print.

Capturing the Sascha Hüttenhain pose

Viewed from the other side

For the 'Pensive' sculpture, I had created a long hair style that would sit comfortably on the woman's collar bones, avoiding another potential printability issue. However, this model is holding her head sideways, so I knew that a long hair style wouldn't work.

Hair is a real problem when using the models in 3D posing software. It usually consists of multiple flat layers combined to give the illusion of depth. This may work for 2D images, but 3D printing demands real depth with solid models and the standard hair add-ons don't work.

One of the most useful techniques I have tried, is to export a hair style as an .OBJ file, import it into Blender and then use the Shrinkwrap modifier on it. It isn't a completely automatic process yet, because it still needs a fair degree of tweaking, but it is much better than attempting to print the original hair model. The finished hair object can be combined with the rest of the body with a Boolean Union join to create a single watertight mesh which the slicing software can cope with.

This model presents several new challenges. The head is tilted sideways; the legs are at a shallower angle; the left elbow is beside the knee, not resting on it; the hands are being held horizontally, one above the other.

All of these factors led me to the conclusion that I would have to turn on automatic support in my slicing software. I would also have to print with ABS plastic because of the way that the support elements will have to break away from the model, leaving the minimum of scarring. I fully expect to have to do some post-processing work after printing this model.

I generally use Slic3r for turning my 3D models into printer instructions, but I don't find it that good at generating support material. Recently, I have had more success with using Kisslicer when I need good support material. Printing models with a small footprint, and this includes support material, is sometimes more reliable if it is laid down on a printed raft, rather than directly onto the printer bed. 

In Kisslicer, I turned on the Raft option and set the Support material density to medium.
I set the model height to 8 cm and the layer height to 0.2 mm, giving a total of 400 discrete layers.

Here is a close-up preview of the printer file rendered by the excellent printer controller software, Repetier-Host, showing the raft (under the feet), the support structures (under the legs) and the individual layers of the model.

Repetier-Host preview render

And here is the 3D printed version.

Printed in grey ABS at 240°C, 0.2 mm layer height, waiting for support material to be removed.

And now, following removal of the support material and surface blemishes.

Removing the support material

Video

Pensive

The 'Sun worship' model took me at least five attempts to get a decent print from, but having finally got it to the point where I was happy with the output quality I decided it was time to move on and try a different design.

There are many factors to consider when designing a model for printing on a 'fused filament' type of 3D printer, like a RepRap or Makerbot. One of the most important ones is the the need for 'support material'.

The printer starts printing the bottom layer onto a flat bed (usually heated to 100°C to make the plastic stick to it) and then proceeds to print each successive layer on top of the one underneath. This works fine for blocks with straight sides and can even cope with sloping sides of up to 45 degrees where each layer projects out a bit further than the one below.

But in real life, things stick out at all sorts of angles and this presents a problem. Just take a look at the middle of your face and imagine how difficult it is to print chins and noses. The printer is squirting hot, runny plastic filament downwards from the nozzle and needs something to extrude the plastic onto. If there is nothing there, the string of plastic will just hang down in mid air instead of following the path that the nozzle is tracing as it draws.

Sun Worship

One way to address this problem is to create support structures which are not part of the sculpture, but are solely intended to provided a surface for the upper layers to land on. 
Another way is ensure that the model has no projecting angles shallower than about 45 degrees.
Look again at the Sun worship model and you will see that all of the limbs, the torso and the head have been carefully posed to ensure that there are no unsupported regions.



I decided that my next project would be a nude sculpture of a woman standing in a pensive pose, one knee slightly bent and arms held close to the chest. 


After working through a number of iterations I arrived at this pose.


Again I spent a considerable amount of time fixing the mouth and eyes and creating a customised hair object.

It turned out that there were several problems with this model. Firstly, the points of the elbows and the chin had nothing beneath them so I decided to add some small support blocks for removal after printing.


In this picture, you can just make out the small disc between the thumbs, as well as the triangular elbow supports.


This should have worked in theory, but in practice the chin support was too thin and failed to print correctly, while the elbow supports ended up too close to the model and caused an unsightly mess.






The other big problem was that my printer can only go up to a height of 13 cm and even using the full print height the model still ended up quite small.
This would have been OK had it not resulted in the fingers becoming too thin to print correctly.
Here is one of the early attempts demonstrating some of these problems.

I made a number of significant changes to try and improve the quality of the print.

• Moved the elbows closer to the abdomen and removed the support blocks
• Repositioned the thumbs under the chin
• Slightly fattened the fingers and thumbs
• Increased the height of the model to 20 cm
• Printed in two halves, with a connecting pin to join them back together

The fingers are little bulky and the seam is still visible but this is straight off the printer, with no post-processing work.

Pensive

Incidentally, you can see the legs of this model being printed in the very first picture of this blog.