Create High Quality HDRI for VFX using RICOH THETA Z1

Create High Quality HDRI for VFX using RICOH THETA Z1 (Part 3/3)

Sample data

Original data including RAW data used in this explanation

DNG, TIFF, each process data download - Size 1GB

Other sample HDRI
Both are based on sRGB development. Compliant with current flow without light source correction.

License CC0

Sample Download


Example of use in Blender

A free addon called HDRI Sun Aligner is useful for aligning the sun on Blender.

Development with Iinear Rec2020

When developed in darktable with Iinear Rec2020, the image is TIFF, but the gamma is exported as an Iinear image. If this is the case, HDR-Merge will be read as sRBG (gamma2.2) in Blender when HDRI is combined, and processing will not be performed properly.


HDR-Merge-master \ build \ blender \

Open with text

img = (img_path) = ‘Linear’

If is described, the image can be forcibly loaded as Iinear and can be processed appropriately.

There is no information on how much color gamut the actual Z1 has as a sensor, and we have not verified it. Since it is HDRI of Rec2020 color space, it is necessary to match the color space read by CG software. It might be easier to use Nuke etc. to convert to ACEScg etc.

Note: RawTherapee has many color profiles such as ACES, and the gamma can be changed arbitrarily.

Bonus ①Stitch with RICOH THETA Stitcher

RICOH official RICOH THETA Stitcher is released as a plug-in for Adobe Lightroom Classic.

The initial settings are difficult to understand, and you can set it so that you can start it from Lightroom by looking at the manual in the installation folder.

I’m thankful if you start it alone, but I can’t. However, you can use it by dragging and dropping TIFF data to RICOH THETA Stitcher.exe with Lightroom running.

The original DNG file is required on the same level. Pressing OK will overwrite the TIFF with the stitched data. 16bitTIFF can also be processed.

The stitches are the cleanest because the official lens correction values ​​are used. However, since numerical values ​​are also automatically input from the image metadata every time, it is necessary to manually correct the images one by one and align the brackets, which is not practical.

Bonus ② Stitch with Nuke

Stitch using Nuke’s Spherical Transform. Spherical Transform doesn’t process BBox, so you can’t blend the joints. This is enough for lighting.

Also, although it has not been verified, it seems to be able to stitch using Cara VR of a paid plug-in (it seems to be integrated with NukeX12). High cost.

Crop of lens> Corrected because the center of the lens is slightly shifted> 190 °, so adjust the scale so that the angle of view is 180 °> Deformation from fisheye lens> Merge

Nukepedia’s free plug-in for Nuke called J_Ops has an HDR tool called HDR merging that can be completed only with Nuke after development, but unfortunately there is no update after Nuke8 support and the latest version Cannot be used with Nuke. There is no similar plug-in like Gizmo.

Naturally HDRI Haven’s HDR-Merge can be assembled by Nuke or Natron as if it were processed by Blender’s composite node in the back.

Especially in THETA, the shooting rules are difficult to shift, so there is no need to pull the value from the metadata every time, so it is easy to create a template.

Bonus ③ Stitch with Natron (Nuke)

I’ve done so far with free software, so I wondered if Natron could manage it. However, there is no node equivalent to Spherical Transform as Nuke.

So STMap can’t be used. You can use either Nuke or Natron.

Original STMap> Read STmap with Hugin and apply only deformation> Adapt STMap with Natron

Of course, the merge process is only blurred with a mask, so image merge processing like Hugin is not done. However, you can get a pretty good result by adjusting the exposure with a little mask.

Although it is not universal, once it is made into a template, the processing is the fastest and becomes 32bit processing, so the drawback of becoming Hugin’s 16bit processing can be solved. The resolution is 7200 x 3600 combined with Hugin.

I feel good with this. That’s why I distribute this data. STMapA, B, Mask, Naron data

THETA Z1 STMap Download

Bonus ④ Correction by Gimp.

The color temperature adjustment in Gimp’s color adjustment was introduced because it was convenient.

It is convenient that you can change the color temperature from any color (during development) to the specified color temperature.

In addition, clone brush and correction brush can be used for both 16bit float and 32bit float, so it is convenient when there is something you want to erase.

1 Like

Hi, if we of the HDRI created in the developing flow of Rec2020, it was found to be able to maintain the color gamut equivalent to DCI-P3.
This is important for use with VFX.
The performance of Z1 can be pulled out to the maximum!

1 Like

Thanks for the update and all your great work. Look forward to more awesome content from you. :slight_smile: :theta::thetadev:

This is a Japanese magazine for the CG and video industry, and I’m writing an article on creating HDRI, which is an update to my blog post.
Describes how to create higher quality HDRIs for scene-linear workflow/ACES using IDT (Input Device Transform).
It is possible to match the color of the backplate with the color of the shot taken with a major cinema camera such as Alexa with IDT.
After the release date, IDT and other data introduced in this magazine will be available for download at “”.

THETA Z1 is also used for the cover and the feature cover.


Sample data is available at
A chapter on comparing renderings, which appears in this magazine, is also included.
Unfortunately, the use of the sample data needs to be seen in this magazine. However, it will also be posted on the web in about a month or two.
Please note that all of them are written in Japanese.

The WEB version is now available.


Bracket shooting plug-in released.

Although there is some camera shake, it is possible to create a handheld HDRI in a crowded place like the scramble crossing in Tokyo Shibuya!


Hi, can anyone share workflow for creating vfx HDRI using Theta X? Authydra doesnt seem compatible with it ( not visible in theta X plugin store)

This thread has some additional information.

This article has more info.

I am only an amateur hobbyist. My settings may not be good. However, you can see use with Luminance HDR (free)

More info from older cameras.

The THETA X is not suitable for creating high-quality HDRIs because it cannot shoot RAW. The Z1 is positioned above the X.
Therefore, our Burst-IBL-Shooter plugin will not support the THETA X.
Another thing is that the API “burst mode” is not available for the X, so similar high speed bracketing is currently not possible.

It is possible to create HDRIs using standard bracketing in JPEG, but JPEG is not suitable for high quality HDRI creation due to the tone mapping applied to JPEG.
Nevertheless, there may be a plug-in in the future to create HDRIs from JPEGs. X has just been released.

Sorry CGSLAB, but “THETA X is not suitable for creating high-quality HDRIs because it cannot shoot RAW” incorrect sentence.

With Theta X harder to create high-quality HDRIs because Theta X only shooting in JPEGs with tonemapped applied. But this only back us to early 2000s when first digital cameras start selling.
That days we also only had jpegs.
Definitely JPEGs are limiting users but biggest issue is not tonemaping but jpegs artifacts and in-camera stitiching (API allow save dual fisheye JPGs but not firmware).
Tonemapping is reversable, and original HDRI papers show how to reverse engineer responsive curve (linearize).
Professional software like PTGui Pro still have option to recompute responsive curve for similar cases.
So sRGB with EOTF can be reversed to linear sRGB and later can be merged to correct HDRIs.

THETA X way more suitable and comfortable to capture HDRIs than slow but bigger sensor Theta X.
CGSLAB burst plugin actually unusable for proper HDRI capture because use automatic exposure control and only limited to 9 brackets.
When Theta X natively support burst capture and can allow 13x exposure brackets. That allow setup 1.3EV distance to capture from 1/16000 to 15sec.
For daytime captures not Theta X nor Theta Z1 is not suitable to capture sunny scenes without ND filters.

And here some examples of high quality HDRIs and Image based renderers made from them.
Dynamc range about 27EV, capture time about 30sec (two captures without and with ND filter)

Or nightime HDRI also around 24EV

Theta X more suitable to capture indoor or night HDRIs due to 1/16000 fastest speed.

I just wish Theta Z2 will have same resolution and speed as X and same 1inch sensor and shutter speeds as Theta Z1.
Than this will made Theta dream machine to capture high quality HDRIs including spectral correct results.


Thanks for posting this information about using the X. There’s another VFX professional that has both the X and the Z1 and uses the X in certain circumstances as well. There’s more information on his workflow below. The reasons he uses the X are similar to your reasons. It’s really fascinating to see the real-world workflow of people building VFX HDR files on set professionally and learn about the reasons for their choices. Thanks.


Great post :wink:

I share with you some observations that I made during my tests…

What I like about the X:

  • the speed
  • the screen
  • the resolution
  • removable batteries

What I don’t like:

  • jpeg only (no RAW)
  • the images produced have a lot of noise (this is particularly visible in real size)
  • presence of chromatic aberrations…

For VFX my preference goes for the Z1 for the following reasons:

  • RAW shoot (no loss of information)
  • quality and size of the sensor (1 inch)
  • possibility to shoot with an aperture of 5.6
  • larger shutter speed range (1/25000 to 60 vs 1/16000 to 60 for the Theta X)

A great link: The Definitive Weta Digital Guide to IBL – fxguide


Thank you for posting the verification.

First of all, since there seems to be a discrepancy between our definition and yours, Our definition of ”High quality” is that the background plate taken by the cinema camera and the CG rendered by HDRI match. This includes matching light source colors, exposure balance, etc. If the HDRI is properly created, the color chart shot with the background plate will match the color chart rendered with the HDRI.

This is made more reliable when used in conjunction with Logoscope’s IDTs.

From left to right: 5D3 rendering results, 5D3 shooting images, THETA Z1 rendering results, THETA Z1 shooting images

All workflows including these IDTs are available free of charge on, although only in Japanese.

Now, let me tell you about the problem of using X.
When using JPEGs, it is not possible to accurately reversible tone mapping first. This is due to factors such as 8-bit tonality, JPG compression, and Gamut compression.
In addition, JPEG images contain some color correction by the camera manufacturer along with tone mapping, which cannot be simply corrected.
Of course, as has been said, it is possible to correct this to some extent by correcting the camera response curve. However, I am also aware that the HDRIs created with this method do not reach the high quality results that were initially shown.
I apologize in advance for the speculation, but it seems to me that your rendering of the CG is actually misaligned when combined with the background plate.

Of course, HDRIs created using these JPEGs can be used for lighting, so there should be no problem in checking the assets or using them in full CG productions.
However, for live-action VFX, it is important to match the plates that were photographed, and this can cause problems.

There seems to be a misunderstanding about this.
Burst-IBL-Shooter is based on automatic compensation and first includes an offset of -5EV that can be set by the API. Based on this, it takes 7 shots bracketed in +/- 3EV steps, so it can properly bracket the range required for HDRI, which is possible under Z1 specifications.

While Z1 specifications do not allow it to capture sunlight and other light sources, it can be used in most of the shooting and studio lighting environments used on many VFX shoots without any real problems.
This has also been confirmed by many VFX production companies.

The THETA X has not yet been released in Japan, and we have yet to see it in action, but we are hoping that an update will allow us to shoot in RAW, or that the next model after the Z1 will allow us to shoot even better.


Burst-IBL-Shooter has been updated to allow the use of a self-timer
Please see the plugin page for details.


Definitely JPEGs and especially stitched JPEGs are adding a lot of troubles.
But i think it possible reach same results with Theta X as Z1.

I love Logoscope work and his knowledge in colors. But sensor spectral response is not a panacea and not always work as planned. (Plus required too expensive software like Nuke to use them)

BTW, as i see there is no perfect match in renders vs photo:

And can you add user defined EV distance? 3EV not recommended to APS-C and FF sensors but on 1inch they are definitely add too many noise. So only way to use Z1 captures is probably use two burst shots with 1EV or 1.3EV steps with Max to 6~15sec shutter speed.

If the results of X using JPEG and the results of Z1 using RAW are the same, please share your method of creation, or compare it with a color chart shot with a different camera like we did, or compare it with a reference to see if the highlights, shadows, reflected colors, etc. match the back plate at the time of compositing and the lighting matches. Only when we can confirm that the lighting matches the backplate at the time of compositing, we can say that it is a high-quality HDRI.

At least, we would be happy to know this, because in the verification we have done, we have encountered quality problems in terms of lighting balance and reflected colors when using JPEGs.

So, going back to the original topic, I said that JPEGs are not capable of producing high quality HDRIs, with the implication that it is difficult to create HDRIs of a quality that can be used in high end VFX with JPEGs.

Sorry about this.
The difference is caused by the shader (roughness, specular) settings in the CG chart, and it is our fault for not adjusting it enough, but I think this difference is within the acceptable margin of error.
In fact, what is remarkable about this image is that the HDRI renderings produced by the full-size SLR 5D3 and Z1 are nearly identical.
This result indicates that the quality of the light is equivalent to the HDRI created by the full-size SLR.

I don’t believe the 3EV step is problematic since the RAW is used. In fact, as per the attached image, the difference in comparison to the 5D3 HDRI is slight. Noise also has little impact when used as a light in HDRI, and we have determined that a stack to remove noise is not necessary in our verification.(However, we do add denoising during the raw photo processing process)
In HDRI, it is more important to be able to accurately reproduce realistic light values than to have a clean, noise-free image.

Our plug-ins are also set up to focus on finishing the shoot faster. This is necessary both because of the need to shoot quickly on a VFX shoot and to minimize the effects of crowd and cloud movement.
If you need a clear, noiseless HDRI shot, which we use in reflections, we too think you need to increase the number of brackets as you have said, or shoot with a SLR.

1 Like

What’s the best way we can help promote your plug-in? Is there any way for you to make money from your plug-in? We can help promote it with things like this blog post and social media.

It’s better for continuous improvement of the plug-in if the developer can make some money in some business model. Happy to help you promote your business if we can.

This site receives sponsorship from RICOH. As part of that, we can try to promote the great work of the developers.

1 Like

I clearly understanding weakness of Theta X comparing to Theta Z1.
My point only about your definition of “High Quality HDRI” that make Theta users confusing.
Bandai Namco proposed good definition to HDRIs that match scene - “True HDRI”. It not put any label about “Quality”.

And if capture don’t have a goal of exact matching of Backplates and HDRIs but captured HDRI must match scene colors and dynamic range even Theta X can be good. Plus 11K downscaled to 8K even allow to use such HDRIs for reflections or even for background.

I also made a quick and rough test for HDRI with xRite.
result HDRI match scene well, and lightings looks similar to captured moment.

And that’s often what 3D artists want from HDRI :wink:

1 Like
1 Like