20180919_1238194196494368087047066.jpg

422 Comments

• 

  ursotak

  01/16/2026 at 9:08 am Reply

  The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.
  Chinook salmon (Oncorhynchus tshawytscha) display remarkable life history diversity, underpinning their ability to adapt to environmental change. Maintaining life history diversity is vital to the resilience and stability of Chinook salmon metapopulations, particularly under changing climates. However, the conditions that promote life history diversity are rapidly disappearing, as anthropogenic forces promote homogenization of habitats and genetic lineages. In this study, we use the highly modified Yuba River in California to understand if distinct genetic lineages and life histories still exist, despite reductions in spawning habitat and hatchery practices that have promoted introgression. There is currently a concerted effort to protect federally listed Central Valley spring-run Chinook salmon populations, given that few wild populations still exist. Despite this, we lack a comprehensive understanding of the genetic and life history diversity of Chinook salmon present in the Yuba River. To understand this diversity, we collected migration timing data and GREB1L genotypes from hook-and-line, acoustic tagging, and carcass surveys of Chinook salmon in the Yuba River between 2009 and 2011. Variation in the GREB1L region of the genome is tightly linked with run timing in Chinook salmon throughout their range, but the relationship between this variation and entry on spawning grounds is little explored in California’s Central Valley. We found that the date Chinook salmon crossed the lowest barrier to Yuba River spawning habitat (Daguerre Point Dam) was tightly correlated with their GREB1L genotype. Importantly, our study confirms that ESA-listed spring-run Chinook salmon are spawning in the Yuba River, promoting a portfolio of life history and genetic diversity, despite the highly compressed habitat. This work highlights the need to identify and protect this life history diversity, especially in heavily impacted systems, to maintain healthy Chinook salmon metapopulations. Without protection, we run the risk of losing the last vestiges of important genetic variation.
  Commonly used image tokenizers produce a 2D grid of spatially arranged tokens. In contrast, so-called 1D image tokenizers represent images as highly compressed one-dimensional sequences of as few as 32 discrete tokens. We find that the high degree of compression achieved by a 1D tokenizer with vector quantization enables image editing and generative capabilities through heuristic manipulation of tokens, demonstrating that even very crude manipulations — such as copying and replacing tokens between latent representations of images — enable fine-grained image editing by transferring appearance and semantic attributes. Motivated by the expressivity of the 1D tokenizer’s latent space, we construct an image generation pipeline leveraging gradient-based test-time optimization of tokens with plug-and-play loss functions such as reconstruction or CLIP similarity. Our approach is demonstrated for inpainting and text-guided image editing use cases, and can generate diverse and realistic samples without requiring training of any generative model.
  Generating realistic images with AI is difficult because images contain hundreds of thousands of pixels with complex relationships. To make this easier, the image generation task is typically split into two steps: first “compress” the image into a smaller set of meaningful pieces called “tokens,” then learn how these tokens relate to each other.Recent advances have created extremely efficient compression methods that can represent an entire image using just 32 small integers. We discovered that these compressed representations actually capture surprisingly rich information about what’s in the image that humans can understand.More importantly, we found that you can edit images by simply manipulating these 32 tokens directly — no complex AI training required. Furthermore, we show that this enables users to define any custom goal or “objective function” for how they want their image to look, and our system can achieve it in just a few seconds without needing to train new models. Our examples demonstrate this approach for various image tasks like text-guided editing, filling in missing parts, and generating new images from text descriptions.
  Your real compression performance will probably depend a lot on the data you are putting in. Is it all geometries? If you have a lot of non-spatial data (or a lot of text attributes for spatial points), then it doesn’t really matter what you do the geometries – you need to find some way to compress that data instead.
  As others have said, I think you are going to struggle to find a format that meets your compression requirements. You would have to create your own custom format, which given your requirement to use commercial software is not going to be viable.
  I think you need to possibly first consider how you can make your data models more efficient, then look at the compression aspects. For example, do you have a lot of repetition of geometry? You could then have a base set of geometry layers with unique IDs and then separate attribute data sets that reference the geometry by ID – that way you can have multiple views of the same geometry serving specific functions. Most decent software packages will then allow you to create joins or relates in order to create the unified view for a layer.
  GML is a good example of a format that supports this kind of relational data model, though being a verbose format file sizes will be large. You can compress GML using gzip compression and can potentially get a 20:1 ratio but then you are relying on the software being able to support compressed GML.
  Regardless, I would urge you to first look at your data models and see where there could be savings to be had. FME from Safe Software is your best bet if you need to start manipulating your data models.
  To achieve that sort of ratio, you could use some sort of lossy compression, but I don’t know of anything that uses it, and although I have a couple of ideas on how one might implement it, it would be far from standard. It would be much much cheaper to kit your server out with a 1TB disk than to spend time and money developing a custom solution.
  You are also confusing data storage with data representation. Your 4th point mentions being able to view the data at different scales, but this is a function of your renderer, not the format per se. Again, a hypothetical lossily compressed file could store data at various resolutions in a sort of LoD structure, but that is likely to increase data size if anything.
  If your data is to be on a server somewhere accessible by mobile applications, you’re far better off using existing tools that have been designed for the purpose. A WFS server (such as GeoServer or MapServer) is ideally suited to this sort of application. The client makes a request for data of a specific area, normally that covered by the screen, and the WFS sends vector data for just that area, so all the heavy lifting is done by the server. It’s then up to the application to render that data. An alternative would be to use the WMS features of MapServer and GeoServer, in which all the rendering is done by the server, and then it sends an image tile to the client. This enables features such as server-side caching of tiles, as well as scale-dependent rendering, with the minimum of work by you. They both read myriad formats, so you can author your data exactly how you like, and store it where you like, and they do all the cool stuff. Quantum GIS also has a WMS server, so you can author and serve data all in the same application.
  This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: -nc/4.0/.
  A 46-year-old man presented with sudden onset of chest pain. He was in cardiogenic shock at arrival. Based on the results of ECG and echocardiogram, he was diagnosed with ST-segment elevation myocardial infarction. Point-of-care ultrasonography (POCUS) did not reveal acute aortic dissection (AAD). During an emergency coronary angiography, aortic dissection was detected and computed tomographic angiography (CTA) revealed Stanford type A AAD with a highly compressed true lumen. Because of this form of aortic dissection, the enlarged false lumen could be potentially misidentified as a normal aorta in POCUS. Although POCUS is useful when AAD is suspected, we should not overestimate its findings and lower the threshold for CTA.
  Acute aortic dissection (AAD) is a fatal disease that presents in the emergency department (ED). However, the symptoms and severity of AAD at the time of presentation vary and are often difficult to diagnose. As AAD sometimes mimics myocardial infarction, emergency physicians (EPs) are faced with a difficult decision. Although point-of-care ultrasonography (POCUS) has proven to help distinguish between these two diseases,1 we experienced a case in which the form of the dissection made it difficult to diagnose using POCUS.
  Point-of-care ultrasonogram. (A) Left parasternal long-axis view of the heart. (B) The descending aorta is posterior to the left ventricle. (C) The suprasternal view. Aortic dissection cannot be identified on any image. Ao, aorta; LA, left atrium; LBV, left brachiocephalic vein; LV, left ventricle.
  He was diagnosed with ST-segment elevation myocardial infarction (STEMI). After the administration of aspirin and prasugrel, an emergency coronary angiography (CAG) was performed. During CAG, a dissection was detected in the ascending aorta. The right coronary artery was obstructed and drug-eluting coronary stents were placed (figure 3). After CAG and percutaneous coronary intervention, computed tomographic angiography (CTA) was performed. It revealed a Stanford type A AAD with a highly compressed true lumen (figures 4 and 5), large intestinal ischemia and left renal infarction. The DeBakey type I aortic dissection extended to the bilateral internal iliac arteries (figure 6). https://drive.google.com/file/d/1r4yGEOqQWsgm88a8WHnqrS4Rsee_QHkL/view
  68cf12514e ursotak

• 

  ursotak

  01/16/2026 at 9:08 am Reply

  The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.
  Chinook salmon (Oncorhynchus tshawytscha) display remarkable life history diversity, underpinning their ability to adapt to environmental change. Maintaining life history diversity is vital to the resilience and stability of Chinook salmon metapopulations, particularly under changing climates. However, the conditions that promote life history diversity are rapidly disappearing, as anthropogenic forces promote homogenization of habitats and genetic lineages. In this study, we use the highly modified Yuba River in California to understand if distinct genetic lineages and life histories still exist, despite reductions in spawning habitat and hatchery practices that have promoted introgression. There is currently a concerted effort to protect federally listed Central Valley spring-run Chinook salmon populations, given that few wild populations still exist. Despite this, we lack a comprehensive understanding of the genetic and life history diversity of Chinook salmon present in the Yuba River. To understand this diversity, we collected migration timing data and GREB1L genotypes from hook-and-line, acoustic tagging, and carcass surveys of Chinook salmon in the Yuba River between 2009 and 2011. Variation in the GREB1L region of the genome is tightly linked with run timing in Chinook salmon throughout their range, but the relationship between this variation and entry on spawning grounds is little explored in California’s Central Valley. We found that the date Chinook salmon crossed the lowest barrier to Yuba River spawning habitat (Daguerre Point Dam) was tightly correlated with their GREB1L genotype. Importantly, our study confirms that ESA-listed spring-run Chinook salmon are spawning in the Yuba River, promoting a portfolio of life history and genetic diversity, despite the highly compressed habitat. This work highlights the need to identify and protect this life history diversity, especially in heavily impacted systems, to maintain healthy Chinook salmon metapopulations. Without protection, we run the risk of losing the last vestiges of important genetic variation.
  Commonly used image tokenizers produce a 2D grid of spatially arranged tokens. In contrast, so-called 1D image tokenizers represent images as highly compressed one-dimensional sequences of as few as 32 discrete tokens. We find that the high degree of compression achieved by a 1D tokenizer with vector quantization enables image editing and generative capabilities through heuristic manipulation of tokens, demonstrating that even very crude manipulations — such as copying and replacing tokens between latent representations of images — enable fine-grained image editing by transferring appearance and semantic attributes. Motivated by the expressivity of the 1D tokenizer’s latent space, we construct an image generation pipeline leveraging gradient-based test-time optimization of tokens with plug-and-play loss functions such as reconstruction or CLIP similarity. Our approach is demonstrated for inpainting and text-guided image editing use cases, and can generate diverse and realistic samples without requiring training of any generative model.
  Generating realistic images with AI is difficult because images contain hundreds of thousands of pixels with complex relationships. To make this easier, the image generation task is typically split into two steps: first “compress” the image into a smaller set of meaningful pieces called “tokens,” then learn how these tokens relate to each other.Recent advances have created extremely efficient compression methods that can represent an entire image using just 32 small integers. We discovered that these compressed representations actually capture surprisingly rich information about what’s in the image that humans can understand.More importantly, we found that you can edit images by simply manipulating these 32 tokens directly — no complex AI training required. Furthermore, we show that this enables users to define any custom goal or “objective function” for how they want their image to look, and our system can achieve it in just a few seconds without needing to train new models. Our examples demonstrate this approach for various image tasks like text-guided editing, filling in missing parts, and generating new images from text descriptions.
  Your real compression performance will probably depend a lot on the data you are putting in. Is it all geometries? If you have a lot of non-spatial data (or a lot of text attributes for spatial points), then it doesn’t really matter what you do the geometries – you need to find some way to compress that data instead.
  As others have said, I think you are going to struggle to find a format that meets your compression requirements. You would have to create your own custom format, which given your requirement to use commercial software is not going to be viable.
  I think you need to possibly first consider how you can make your data models more efficient, then look at the compression aspects. For example, do you have a lot of repetition of geometry? You could then have a base set of geometry layers with unique IDs and then separate attribute data sets that reference the geometry by ID – that way you can have multiple views of the same geometry serving specific functions. Most decent software packages will then allow you to create joins or relates in order to create the unified view for a layer.
  GML is a good example of a format that supports this kind of relational data model, though being a verbose format file sizes will be large. You can compress GML using gzip compression and can potentially get a 20:1 ratio but then you are relying on the software being able to support compressed GML.
  Regardless, I would urge you to first look at your data models and see where there could be savings to be had. FME from Safe Software is your best bet if you need to start manipulating your data models.
  To achieve that sort of ratio, you could use some sort of lossy compression, but I don’t know of anything that uses it, and although I have a couple of ideas on how one might implement it, it would be far from standard. It would be much much cheaper to kit your server out with a 1TB disk than to spend time and money developing a custom solution.
  You are also confusing data storage with data representation. Your 4th point mentions being able to view the data at different scales, but this is a function of your renderer, not the format per se. Again, a hypothetical lossily compressed file could store data at various resolutions in a sort of LoD structure, but that is likely to increase data size if anything.
  If your data is to be on a server somewhere accessible by mobile applications, you’re far better off using existing tools that have been designed for the purpose. A WFS server (such as GeoServer or MapServer) is ideally suited to this sort of application. The client makes a request for data of a specific area, normally that covered by the screen, and the WFS sends vector data for just that area, so all the heavy lifting is done by the server. It’s then up to the application to render that data. An alternative would be to use the WMS features of MapServer and GeoServer, in which all the rendering is done by the server, and then it sends an image tile to the client. This enables features such as server-side caching of tiles, as well as scale-dependent rendering, with the minimum of work by you. They both read myriad formats, so you can author your data exactly how you like, and store it where you like, and they do all the cool stuff. Quantum GIS also has a WMS server, so you can author and serve data all in the same application.
  This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: -nc/4.0/.
  A 46-year-old man presented with sudden onset of chest pain. He was in cardiogenic shock at arrival. Based on the results of ECG and echocardiogram, he was diagnosed with ST-segment elevation myocardial infarction. Point-of-care ultrasonography (POCUS) did not reveal acute aortic dissection (AAD). During an emergency coronary angiography, aortic dissection was detected and computed tomographic angiography (CTA) revealed Stanford type A AAD with a highly compressed true lumen. Because of this form of aortic dissection, the enlarged false lumen could be potentially misidentified as a normal aorta in POCUS. Although POCUS is useful when AAD is suspected, we should not overestimate its findings and lower the threshold for CTA.
  Acute aortic dissection (AAD) is a fatal disease that presents in the emergency department (ED). However, the symptoms and severity of AAD at the time of presentation vary and are often difficult to diagnose. As AAD sometimes mimics myocardial infarction, emergency physicians (EPs) are faced with a difficult decision. Although point-of-care ultrasonography (POCUS) has proven to help distinguish between these two diseases,1 we experienced a case in which the form of the dissection made it difficult to diagnose using POCUS.
  Point-of-care ultrasonogram. (A) Left parasternal long-axis view of the heart. (B) The descending aorta is posterior to the left ventricle. (C) The suprasternal view. Aortic dissection cannot be identified on any image. Ao, aorta; LA, left atrium; LBV, left brachiocephalic vein; LV, left ventricle.
  He was diagnosed with ST-segment elevation myocardial infarction (STEMI). After the administration of aspirin and prasugrel, an emergency coronary angiography (CAG) was performed. During CAG, a dissection was detected in the ascending aorta. The right coronary artery was obstructed and drug-eluting coronary stents were placed (figure 3). After CAG and percutaneous coronary intervention, computed tomographic angiography (CTA) was performed. It revealed a Stanford type A AAD with a highly compressed true lumen (figures 4 and 5), large intestinal ischemia and left renal infarction. The DeBakey type I aortic dissection extended to the bilateral internal iliac arteries (figure 6). https://drive.google.com/file/d/1r4yGEOqQWsgm88a8WHnqrS4Rsee_QHkL/view
  68cf12514e ursotak