Hi, @codetricity,
I successfully got “Very Good” result with PTGui pro ver.11. It would be better to convert the images into Single-fisheye images.(Crop & Rotate) 4 shots 90 degrees apart definitely will improve image quality.
You can see the stitched images in Google Street View.
https://goo.gl/maps/SCVwcRAUwBq
Enjoy!
Toyo
This is a great add. Thanks for sharing it.
How are you cropping and rotating the images? Are you using something like Photoshop? Or, are you using something like ImageMagick in an batch processing script?
I’m going to add this to the Plug-in Developer Guide that we’ll be using in the August 30 microconference in San Francisco. I will provide attribution to you.
I’m also trying to learn more about HDRi prior to the microconference. One thing I learned is that HDRi files are stored in one of three formats: Radiance RGBE, OpenEXR, or Floating point TIFF/PSD.
Will now stitch each individual image and then attempt to combine.
Ichi’s mobile app can stitch all 7 images with a single batch process.
Saving in OpenEXR format using free Picturenaut
Single HDRi file is now 96 MB in size.
Imported 7 images into Photoshop
viewed in Ricoh THETA Desktop Application
Drag EXR file onto Unity.
Create a new material.
Texture Shape: Cube
Unity recommends using Specular Convolution, but it will blur the skybox.
With Specular Convolution
I just used GIMP to crop and rotate the images. Anyway,the batch processing script will be needed.;lol.
I’m also trying 7 image dual-fisheye shooting. I downloaded the ZIP file from your GitHub site. Is it app-debug.apk? And can I install it in the same camera the same way as 3 HDR version?
Thanks,
Toyo
yes, it is app-debug.apk. You need to install it with adb install -r app-debug.apk
Make sure you uninstall Plugin Application first.
The following will crash the plug-in
Hi, @codetricity
Installed 7images Plugin.
App permissions OK.
I have another problem?
The first image of 7 was the value of 0.0 step. (I had two images with same value) No image of +2.0 step. I shot these images in the room because of the rainy day. Tomorrow, I’m going to test 7images plugin ouside.
Thanks,
Toyo
I believe I fixed the problem of duplicate exposure settings. I can now get 7 images all with different exposure settings. I uploaded to GitHub in a “dev” branch.
Test 2
I updated the apk on the dev branch.
https://github.com/codetricity/theta-7-image-dual-fisheye/tree/dev/apk
It’s also available below.
app-debug.apk (1.5 MB)
Look forward to your test shots. I’m actually finding the HDR techniques useful, making the project more real-world than just a demo for the workshop. Although the THETA V HDR from Ricoh is better, I didn’t like to wait to take the pictures.
The Ricoh THETA Camera API can adjust the shutter speed from 1/25000 to 60. This is a much wider range than the exposure compensation. I’m thinking of altering the workshop code to use shutter speed and create a 9 image HDR with a wider range of values. Does anyone know if this is a good idea or not?
Hi, @codetricity Thank you for quick update. I also can get 7 images +2.0 / -2.0 EV.
Result of PTGui Stitching is “very good” again. I sent my pts file to the support of PTGui.
Coming soon.
Toyo.
Thanks for your help testing the sample code that we’re using in the August microconference.
That’s a cool shot. I used to live in Japan many years ago and it’s nice to get these images.
I’d still like to learn the workflow that you’re using for PTGui. Although I can stitch the images, I’m not getting a good result. In the screenshot above, it looks like you’re using four dual-fisheye images. Is that correct?
Yes, it is correct. I always shoot a pair of Dual-Fisheye Images to get 4 shots 90 degrees apart. This is my challenge. I believe it can improve image quality of THETA V.
Are you rotating the physical camera in the tripod to get the image rotation?
Thanks again for your help with PTGui. I will try again soon with your technique. You achieved a really nice stitch line That’s a cool neighborhood temple. natsukashii.
Also, thanks for identifying bugs in the sample code that we’re preparing for the workshop.
I made changes to the original 3 image sample code:
app-debug.apk (1.5 MB)
I believe that an improved version from Ichi Hirota will be in the THETA store soon.
In the meantime, I’m going to continue to work this small example for the workshop as the code base is easy to work with.
Added new branch to the repository for 12 image bracketing using shutter speed to get exposure times of 1/25000 seconds to 60 seconds. This version starts into multi-image bracket shooting by default.
app-debug.apk (1.5 MB)
Using dual-fisheye plug-in to reduce shot time. There is a delay of approximately 1 second between shots. Example settings with 12 bracket shooting shows full range of shutter speed from 1/25000 to 60 seconds.
The original plug-in used exposure compensation.
This version uses shutter speed to achieve a wider exposure range.
You can switch this to 9 image bracket shooting or adjust the time spacing between brackets in the code.
The plug-in will boot to multi-bracket mode by default. The original plug-in started in single-shot mode.
Batch processed all images after the shoot to convert from dual-fisheye to equirectangular.
Using HDR Pro.
File size will increase from 5.35 MB to 49.9 MB due to luminosity information on all images.
Hi, @codetricity and @jcasman
This is the latest work on Dual-Fisheye Image(7HDR) with PTGui and Pano2VR. I can constantly get “Very Good” result on PTGui stitching process. But some of small stitch errors usually remain just on the seam line.
“Nothing is perfect”
I removed them with using Pano2VR patching tool. This is great.
I would like to test the following apk.
Modified Exposure Compensation to use 5 images from with values of: -2.0, -1.0, 0.0, 1.0, 2.0
Thanks,
Toyo
Here’s a screenshot in Photoshop. The apk and link to a new GitHub branch is at the bottom of this post. Do you prefer the plug-in to start in 5-image bracketing mode first or single-shot mode first? Currently, it defaults to single-shot mode, but it is easy to change the default mode.
The Exif data of the top image indicates that exposure compensation is +2
The 2nd image is +1.
I don’t really understand how exposure compensation works. Do I need to set the shutter speed to a constant value and then apply exposure compensation?
I’ve changed the name of this version of the plug-in to Juantonto Fisheye. This is to make it easier to save different versions of the plug-in with different settings. It’s also to illustrate that the plug-ins can be customized for specific uses.
If you use Vysor to set the permissions, make sure you open up the correct app. It’s now called Juantonto Fisheye. It uses a different icon.
You must enable permissions for camera and storage.
GitHub branch for juantonto version
juantonto-5-fisheye.apk (1.6 MB)
Changes are described for developer education.
Right click on app in Project view.
Hi, @codetricity
I’ve tried Juantonto Fisheye version and published 5 images to Google Street View through PANO2VR.
It is noisy in the dark area. Can you set ISO50 with 5 steps ±2.0 EV?
Thanks,
Toyo
Thanks for testing this. Great feedback and nice shot.
I’m an inexperienced photographer and am learning about photography through your posts and questions. Thank you.
I attempted to make some good images with low noise in dark conditions. Unfortunately, I wasn’t able to get good settings. I’m going to bring this up at a workshop we have tomorrow and hopefully someone at the workshop will be able to figure out some good settings.
The plug-in below has these values for exposure time:
I was hoping that a wide range would resolve the problem, but I did not get results.
app-debug.apk (1.6 MB)
Dual Fisheye plugin has come to Plugin Store! Here it is.
This is the latest dual fisheye shooting. It was very sunny day before typhoon.
Thanks for your support.
Toyo
PS. After installed Dual fisheye plugin from Plugin Store.
pic.1
pic.2
pic.3
This is awesome. Are you doing anything special to embed the Google Maps image into this forum? It’s not embedding when I put an image in.
BTW, these are some great shots that you are taking.
Nothing special to do so. I just paste the shared URL in the box.
Yeah! Thanks, @codetricity
Next, show me your shots.
