Photomatix Question

<p>I recently sent this question to the support folks at Photomatix. I would be interested in your opinions, as well.<br>

I am a Real Estate Photographer. I shoot homes for use on<br />the internet and brochures for real estate agents. I process all of my exterior<br />shots in HDR with Photomatix Pro 64 Bit. I feel that I have become fairly competent in its<br />use. However, I still struggle with maintaining sharpness and clarity. 99% of<br />my exterior shots are on a very sunny day at approximately noon. I shoot with a<br />Nikon D300 and D3S and a Nikon 24-70mm 2.8 lens in Aperture Priority mode. I<br />shoot in the high speed continuous mode with AEB set for 5 shots in one stop<br />increments. Currently I set my aperture at f16. At ISO 200, this will<br />give me a shutter speed of 1/125. At -2 the shutter speed will be 1/500 and +2<br />will be 1/30. I process the files in Photomatix Pro 64 bit. At this setting I<br />have to do extensive sharpening with Photoshop CS3. I have run test shots at f8 and found that the<br />sharpness actually decreases from the f16 setting. It would seem to me that it<br />should be the other way around. At f8, the shutter speed increases to<br />1/400 and the -2 shot increase to 1/1600. If I take what should be the sweet<br />spot of the lens, f5.6 to f8 and combine it with the higher shutter speed,<br />shouldn't the result in Photomatix be sharper with more clarity? </p>

<p >.<br>

</p>

<p > <br>

</p>

<p > </p>

<p>Sounds to me like the problem isn't the sharpness of a given frame, but rather the movement between exposures. Rather than blasting away at high sustained frame rates, just slow down ... and use a tripod.</p>

<p>Matt - I do in fact use a tripod. I also use a radio trigger to activate the camera. Camera movement is not an issue. I shoot at high frame rates to avoid as much movememt at possible during conscecutive shots. Leaves on trees, flags waving, etc.. Could the shooting at a high frame rate cause camera movement that could be detected at the slower shutter speeds?</p>

<p>Can you post an image that exhibits the problem? It would be best if you post the full frame as well as a 100% crop.</p>

<p>Tom M</p>

<p>Pat there are two options for aligning your exposures, from memory one is horizontal shifts and the other matching features, have you tried both?</p>

<p>Just want to add, convert your files to Tiffs before importing to Photomatix and see an improvement in noise levels, this is the one time where RAW comes second.</p>

<p> </p>

<p>High-speed shooting on a tripod could very well be part of the problem - there really is no reason for it though. Photomatix ghosting algorithm has become very good - and you can help it along manually to eliminate motion artifacts. Also, what setting are you using to align your images (?) - this might well be where the crux of the problem is.</p>

<p><img src="https://picasaweb.google.com/pcassity53/PhotoNet?authuser=0&feat=directlink" alt="" />Dieter - As far as alignment is concerned, I am using 'correcting horizontal and vertical shifts'. However, motion artifacts don't appear to be the problem. Here are the images. 4512 is at f16 and 4517 is f8. I have done no other processing other than the Photomatix conversion. No sharpening has been applied. These were Nikon NEF files from the D300 with the 24-70mm 2.8 lens.</p>

<table >

<tbody>

<tr>

<td align="center"><a href="https://picasaweb.google.com/pcassity53/PhotoNet?authuser=0&feat=embedwebsite"><img src="https://lh6.googleusercontent.com/-KuElCB1z6UM/ThY_hnxqQZE/AAAAAAAApPA/9HaPLpTLzgc/s160-c/PhotoNet.jpg" alt="" width="160" height="160" /></a></td>

</tr>

<tr>

<td ><a href="https://picasaweb.google.com/pcassity53/PhotoNet?authuser=0&feat=embedwebsite">Photo.net</a></td>

</tr>

</tbody>

</table>

<p>To me the f/16 crop look less sharp than the f/8 one - diffraction is very likely playing a role in softening the image at f/16. AFAIK, the 24-70 is also not a lens that is at its best at or near infinity - it appears to be optimized for event shooting and their closer distances.<br>

Now, another question - do you do CA correction. There is a bit visible in the crops - and the correction will shift the three (R,G,B) images relative to each other, which can cause softening due to misalignment as well if the CA correction isn't optimized properly. Also, for optimal quality, you may need to use another RAW converter (NX2 or ACR) and then run the converted images through photomatix; this generates a lot more work though.</p>

<p>I don't see much issue unless you are looking at the zoomed in crop... and pixel peeping will always result in anxiety over sharpness.</p>

<p>There's nothing wrong with your images when viewed the way any normal person would look at them. I can only suggest a few things:</p>

 

<ol>

<li>Get a sturdier tripod or get a sandbag that you can attach to the center pillar of your current tripod to reduce shake.</li>

<li>Slow down the shutter release...a D3s in Ch will shake many tripods...see #1</li>

<li>Stop Pixel peeping</li>

<li>Give up on Photomatix and try <a href="http://www.niksoftware.com/hdrefexpro/usa/entry.php">Nik Software's HDR Efex Pro</a>. I find it to be much better at aligning images and realistic results than Photomatix, (although yours look pretty darn good)</li>

</ol>

<p>Hope this helps<br />RS</p>

<p>I shoot most of my HDR photos using my D3 or now my D3s, handheld with a 17-35 2.8 wide angle at around 5.6 or 8. I shoot on my fastest motor setting, and use photomatrix light. I do not have problems with aligning the 5 photos when processing. If you look at my HDR folder here in Photo.net you will see what I am talking about. I do shoot Raw and then work on the photos in Photoshop to sharpen. I find photomatrix to be very good at aligning the photos together. I recently had a show and many of the photos were blown up to 16x20, and 20x24, still keeping the sharpness. </p><div></div>

<p>(1) Are you using live view to acquire focus? </p>

<p>(2) Have you checked sharpness on the individual shots at 100% viewing? </p>

<p>(3) Are you using NEFs as input to Photomatix? These are not very high quality converters, only there for convenience. Best practice is to capture to 16-bit TIF using the /neutral/ picture control or the /linear/ setting in CaptureOne. Use those as input to Photomatix. You want the most linear data you can get.</p>

<p>(4) The D3/s has a very aggressive AA filter, and in most converters, there is a little bit of blur in everything. Some people use deconvolution to sharpen after capture. It is surprising how well images respond to this. </p>

<p>(5) You will see some diffraction at f/16. Remember, you are stacking up on the diffraction too when you're doing HDR!</p>

<p>Luke</p>

<p>Thanks for all the great responses. These shots were at 34mm. Is it suggested that I convert all files to 16 bit TIF before importing to Photomatix? Is there any advantage to instead shooting jpgs instead of Raw, since I am doing very little (if any) processing of the NEF file before it is imported to Photomatix? Post processing time is a major consideration for me. Processing a jpeg in Photomatix is certainly much quicker than a NEF file. However, I want the best possible result from Photomatix that I can get, before I spending any additional time sharpening, etc..</p>

<p>HDR is a method of supersampling that wants good clean /linear/ data so it can accurately normalize and average. I recommend very strongly against the use of JPGs, which throw away 4 or so bits. I also recommend very strongly against the use of any sweetening whatsoever -- no tone curves, saturation boost, or anything -- until you have a tonemapped image. All of those things falsely skew the data. Suggest using CaptureOne, and the "linear" tone curve. Make sure the WB is the same on all images, then batch capture to 16-bit TIF. Very quick work. Use those as input to Photomatix. Avoid using the raw converters in Photomatix. </p>

<p>Michael Ging: Try increasing light smoothing settings during tonemapping to get rid of those halos.</p>

<p>I would recommend RAW over TIFF, and either one is miles ahead of JPEG.</p>

<p>What size are you printing? Or is this for web-display only? I wouldn't worry about sharpness before processing them and adding a sharpening at the very end. Then print. Examine the print. Is it sufficiently sharp?</p>

<p>Eric</p>

<p>Eric - Photomatix does not exploit the information inherent in the RAW files, and in fact, it's converters, provided for convenience only, are not up to the quality of the commercial converters. It is much better to use 16-bit TIFs made with a good converter.</p>

<p>Luke - Since my output is only for internet use, do you see any advantage to converting to a 16 bit tiff instead of 8 bit tiff before processing with photomatix?<br>

Eric - All of my sharpening is done at the very end of my workflow.</p>

<p>The way I'd see it, yes. In 30 seconds, CaptureOne can capture 6 12MP D3 files as 16-bit TIFs on my setup. This requires no settings except for setting the "linear" tone curve on all images. So there is no need to keep the 16-bit TIFs around unless you have a use for them after the job is delivered. The .HDR files with 32-bits per channel require some time and effort to make and are archival. Best to make them as good as you can, because they are very very good datasets. Future tonemapping tools will outperform today's on the same data.</p>

<p>In technical terms, I feel that you will see a difference in the quality of the shadows, but I can't point to a comparative example that would tell you how much. Think of it this way: The HDR pixel value is computed as an average of the (normalized) samples and rendered as a floating point average. If you use truncated samples, then some pixel values are computed on an average of 2 samples instead of 3 samples, and some are computed on an average of 3 instead of 4 samples. This is on the assumption that 10 bits are "good" out of a 16-bit sample. In fact, the newer sensors are "good" out to 12. </p>

Pat, Photomatix is generally just an average RAW converter so I wouldn't feed it NEF files. Perhaps its greatest shortcoming with RAW, at least for the way I work, is that it renders out bastard pixel dimensions that don't correspond to the output of either ACR or NX2. In post (Photoshop), the odd sizing makes it impossible to blend an original frame from your bracketed set with the output from PM -- something I do often for various reasons.<br><br>

 

With regard to image clarity or sharpness, PM's alignment is hit and miss and may be an issue. Try merge to HDR in Photoshop and save out an openexr 32 bit file for tonemapping in PM and see if that improves your output. Before you do that, examine your individual frames and make sure they are all sharp. If one is the least bit fuzzy, toss it out..<br><br>

 

Outputting tiffs for use in PM also has merit if for no other reason than you avoid letting PM handle the RAW files. This is my normal workflow..<br><br>

 

Running your images through another piece of software is going to add time and complexity, but if it improves your images noticeably, what other option do you have?

<p>Howard and Luke - Since I do further processing of the image after being processed, do you also suggest that I save as a Tiff file in Photomatix?</p>

I think I would, given that you're also processing for print usage. Whether to maintain a 16bit workflow is up to you. I archive everything in 16bit (my RAW files) but I only carry that so far into my workflow.<br><br>

 

I'd avoid jpegs except as output files. You can work with them and you can even resave them numerous times, but quality loss eventually does creep in and it's really not worth it. Especially since improving your quality is why you came here in the first place.

<p>I suggest tonemapping and saving as a 16-bit TIF for the final. The 16-bit file will give you a superior reduction to web size. I just think a 16-bit work flow is the best for all professional work.</p>

Luke and Howard - Today I tested converting all raw files to 16 bit tiffs with NX2 before importing to Photomatix. I also

made sure that picture control was set to neutral with no sharpening before converting to tiffs. The result from PM was

outstanding! I then saved the tone mapped file as a 16 bit tiff and opened in photoshop cs3 for final processing. The

difference in clarity and sharpnes in the final output was nothing short of amazing! Thanks to both of you for your

patience and thorough explanations.

Pat, that's great news. And thanks for reporting back with your results. It helps to know that the advice given was useful.

<p>Pat, glad to hear it. I went back and looked at your original images on Picasa, and I think the only other comment that I'd make is that you are getting some halo artifacts that are most visible around the trees, but occur elsewhere. [see the variations in the brightness on the sky surrounding the trees, which are due to haloing during the detail enhancement process.] I might recommend a higher setting for light smoothing. For jobs like these, I usually use a low 'strength' setting around 30-40 during tonemapping. Your taste may vary.</p>