A Medieval Discovery at Gulval Church, Cornwall

Gulval Cross-BaseJust to the left of the south porch of Gulval Church, near Penzance in Cornwall, lies a large block of granite. I first came across it on a visit to the churchyard in 2012. However, it really piqued my interest a couple of months ago after reading a short report in Newsletter 132 of the Cornwall Archaeological Society. The article, entitled ‘An unusual stone at Gulval Church‘, by Professor Mick Aston, Teresa Hall, Professor Rosemary Cramp, Ann Preston-Jones and Andrew Langdon, highlighted that the block was a medieval cross-base, and was decorated on all four sides. Aside from the south, pictured here, the images on the other sides were very difficult to see at all. Ann and Andrew visited the stone at night “armed with car batteries, a mirror, two spotlights, a tripod, head torches, and a camera” to capture raking light photos of the decorations to see if controlled shadows could be used to interpret the decorations.

The results were good, producing a set of shadowy atmospheric photos of each side. Professor Cramp began to interpret the scenes from these photos but noted that, while useful, the photos only provided a single fixed view, and she had not seen the cross-base with her own eyes. It was thus difficult to say for sure what was depicted on this rough stone.

It was shortly after receiving the newsletter that I also heard the sad news of Mick Aston’s death. Doubtless that Mick would have had a continued interest in the cross-base, and given that I only live a few miles from the church I became determined to use the tools available to me to record the cross-base and see if I could help with the interpretation.

On 8 June 2013 I visited Gulval Church and recorded in high resolution 3D each side of the cross-base using a photogrammetric approach. Taking a series of over a hundred overlapping 18 megapixel scale photos at a fixed focal length, I covered the whole block.  It was quite a tricky subject to tackle, as there is very limited clearance between the cross-base and two of the church walls.

Over the next few weeks, I began to process the images into 3D point clouds, and from there to solid 3D meshes. Afterwards, I subjected each side to a series of techniques which I have used to analyse prehistoric rock art and inscriptions in the past. The results were fantastic.

What is depicted on the cross-base?

Thanks to this analysis, I believe that we now know what is shown on the cross-base at Gulval. The existing interpretations, as laid out in the CAS Newsletter, didn’t match up with what I have found. The computer-generated images, shown below (and may others that I have created using false colour representations of depth), have been able to provide a slightly clearer representation of the granite carvings, and, I believe, help perceive them in a different way. I can now see many of the details in the original photos now that I know what I am looking for.

Rather than sit on this discovery for ages until everything is just right, I have decided to post my initial findings here. It does not represent a full description of the findings and condition of the cross-base and its imagery, and it builds upon the work already done.

Firstly, I must thank my friend, Nick Ford, for putting me onto the iconography – I believe that he spotted what was depicted in the first image I showed to him straight away, and knew what the others would reveal. It led to an exciting evening searching for comparative images, and further enhancing them. And without the original report in CAS Newsletter 132 (now online) I wouldn’t have got this point at all.

The cross-base depicts the four Evangelists, Matthew, Mark, Luke and John, as the four living creatures described in Ezekiel 10:14 and Rev 4:6-7

Revelation 4:6-7

New International Version (NIV)

6 Also in front of the throne there was what looked like a sea of glass, clear as crystal.

In the center, around the throne, were four living creatures, and they were covered with eyes, in front and in back. 7 The first living creature was like a lion, the second was like an ox, the third had a face like a man, the fourth was like a flying eagle.

The Gulval Four Evangelists

The south side of the stone depicts St Matthew. He is symbolised as a winged man or an angel, and depicted here with a halo, holding a book (Gospels or Book of Life?) with “MT” inscribed on it. The lower part is damaged. This image shows the details picked out using an ambient occlusion filter.

On the west, facing outwards, is St Mark. He is shown kneeling, with haloed lion head. His book, held in his left hand, has “M” on it. It is possible that he is giving the sign of benediction with his right. The folds of his robes are nicely visible.

On the north, facing the church proper, is Luke, depicted here seated with haloed calf head . His book, held in a rather large left hand, has “LS” on it (Lucas). Luke is mainly depicted as an ox, but it isn’t unknown for him to be shown as a calf.

On the east, facing the porch wall, perhaps the most badly eroded of them all, is St John, depicted as the eagle. His robes, especially the folds around his legs, are clearly visible.  It is difficult to discern the presence of any halo. He is holding a book in his right hand, with what looks to be an “H” on it (not easily visible in this image – only range-colouring a very small area shows this). It would be very neat if it said “IH” – Iohan, but archaeology isn’t always a neat discipline. Further work needs to be done here.

We must remember that the stone block onto which these images are carved is a course Lands End granite. Some of the quartz crystals are several centimeters long – fine detail just isn’t possible – and granite can crumble when exposed to the elements.

A Tetramorphic Cross-Base?

So there we have it. Starting on the south side of the stone, if you run the sequence clockwise (South, West, North, East), runs the familiar sequence of Matthew, Mark, Luke and John. Matthew, the best preserved of all, was positioned facing outwards, and the shield-like shape of the carving has, according to some, become thought to be a coat of arms. Now that we know what is really shown here, perhaps we can even call it a tetramorphic cross base, bringing together the symbols of the Four Evangelists together into a single object. One thing is for certain, however, and that is a lot more work needs to be done on the Gulval cross-base, in terms of research, verification, and recording techniques. Reflectance Transformation Imaging (RTI) will certainly be the next approach for further enhancing the panels. Comparative examples need to be sought.

This could have a big impact upon our understanding of Gulval, and indeed have ramifications on how we understand the history of this part of Cornwall in the early Middle Ages.

The authors of the original article in the CAS Newsletter have had copies of the images and my findings in advance of this blog post. Luckily, when I sent my findings to Ann Preston-Jones, she was pleased, as she felt that these were the Evangelists, but lacked firm evidence to back it up. I trust that my arguments, and the images above, will help to win around any doubters.

I have been asked to add the finished images of the Evangelists to the forthcoming Corpus of Early Cornish Sculpture, to which I am honoured to be able to make a contribution. The vicar of Gulval Church is adding the images to his parish newsletter. Who knows, even a sermon may come of it!

Further images and information will be added to subsequent blog posts, and linked from here, so do bookmark this page and check back from time to time.

And now for the plug!

As an independent archaeologist, I am of course pleased to offer my professional services. Please contact me with any enquiries.

Below is a series of images of St John (Eagle) from the east side of the cross-base. The 3D data has been filtered in a number of ways to extract the detail of the carving.

East elevation of a medieval cross base at Gulval, Penzance, depicting St John the Evangelist as an eagle (see Rev 4:6-7 & Ezekiel 10:14). Captured using photogrammetric methods and processed using different filters to interpret the carving.

Google SketchUp 7, 3D export, and Vue 7 Infinite

[Update March 2010] In an update to Sketchup 7, you can now just go to File > Export > 3D Model and choose ‘COLLADA File (*.dae)’ as an option. This post remains as the information about extracting a KMZ file could be useful to some.

Google have just announced the release of SketchUp 7. SketchUp is a wonderfully simple 3D modelling package, often used to populate Google Earth with 3D models of famous buildings. SketchUp Pro is the paid-for ‘grown up’ version of SketchUp, allowing, amongst a range of features, 3D export of models in a variety of formats.

The free version of SketchUp only allows models to be saved in the proprietary .skp file format, and export to kml for inclusion in Google Earth. However, I noticed on the version comparison (Why go Pro) page, that listed in the 2D export feature list, that Collada (a 3D interchange format) export was supported. It’s strange to see it listed as a 2D file format, but there you go. This is exciting, as it means that the free version of SketchUp would be more usable to me (I can’t afford the $499 for the Pro version).

Sketchup Collada export

So, I downloaded SketchUp, and made a quick box model. I can do more complex models, honest!

Simple model in Sketchup 7

The next step would be to export as Collada. Since it was listed as a 2D export format, I looked for it in the 2D export menu, but JPG, PNG and TIFF were the only options. So I checked the 3D export menu, which only listed KML and a tempting link to upgrade to Pro. The help menu didn’t seem to mention Collada export either. Initially, I put this down to an error by Google, and that this was indeed a Pro feature that had slipped into the free version’s feature list.

However, rarely one to give up, I decided to export the model as KML and see what I could do with it. I noticed that KML exports were KMZ files (a compressed file containing geometry and textures). On my Mac, I convinced Stuffit 10 to unzip the KMZ file to my desktop.

unzipped KMZ file

Stuffit created a directory containing the unzipped files. At first this just looked like it was a kml file and materials, but a quick look in the ‘models’ directory revealed a file with the extension .dae – which is the extension used by Collada files. So, through a rather backdoor method, it does indeed export Collada files. Which also means that Google Earth will read Collada files, which could lead to some interesting possibilities.

The next step is to test the exported file. I’m lucky enough to have a copy of E-On Software‘s Vue 7 Infinite at work, which supports the import of Collada files. Windows users of Vue are lucky enough to have a native .skp import, but it’s something us Mac users have to go without. Collada export neatly solves this problem.

The exported file in Vue

It wasn’t that straightforward in Vue, however. The imported Collada file appeared to have no texture applied to it. A look in Vue’s material inspector revealed that for some reason the material had 100% transparency. Setting this back to 0% transparency showed my SketchUp-designed object, which I was able to render. Success!

While SketchUp Pro offers a lot more functionality, as well as plugins, the free version is still very useful for simple modelling tasks. Being able to export in 3D from the free version is a definite boon, and my use of Sketchup will definitely increase as a result.

I hope that this is of use to 3D artists out there – feel free to leave comments if you have any ideas on how to streamline the process.

I’m featured on a 3D software company website


E-On software, developers of Vue 6 Infinite, my 3D package of choice for landscape visualisation, are running a feature on my recent work. It’s quite nice to have been picked by them to show off what the software can do.

But I am left with a little sour taste in my mouth. The latest “update” for Vue 6 Infinite breaks network rendering on the Mac. And what am I currently working on? A large scale animation of a digital elevation model (DEM) of a world heritage site. What do I need? A render farm of lots of computers. Great.

So I’ve rolled the software back to its previous state, and net rendering works again. Or so it seemed. Network rendering using the “HyperVue” and “RenderCow” elements of Vue are shockingly unreliable. I’ve battled with it for years, and it seems to actually be a lot worse than it ever used to be.

Over the last week or so, every render but one (say 9 out of 10) have failed. Either the RenderCow clients have crashed, or the whole thing has gradually just… stopped. Without finishing. Maybe 300 out of a 1000 frames are rendered. Next try, it’ll do 162. The RenderCows just permanently get stuck, one by one.

I’ve got support tickets open, but E-On are going to have to do something quite major to fix this.

Unfortunately, even though I’m now being used to advertise their software, I can’t actually recommend it if you need to do any large scale animations, with real deadlines, if you’re a Mac user. It’s great if you want to use it for a hobby, but you can’t rely upon it.

I have a complete love/hate relationship with Vue. It does produce beautiful images that are incredibly difficult to achieve using other software, and it’s quite simple to use, but it’s just so darned unreliable.

There’s now a Personal Learning Edition version of Vue 6 Infinite, if you’re curious about it.

Google Metaverse?

GigaOM reports that more evidence is turning up that Google are developing their own metaverse, i.e. a virtual 3D world like Second Life.

Google themselves have said nothing about it themselves, so this is pure rumour/wishful thinking.

What with Google’s acquisition of SketchUp and Keyhole which demonstrates their keenness in 3D, it wouldn’t surprise me if a Metaverse is up their sleeves. We shall have to wait, watch, and see…